{ "schema_version": 2, "meta": { "build_id": "DATABRICKS-GENIE-SWARM", "slug": "databricks-genie-swarm", "title": "Build Your Own AI Swarm in Databricks with Genie", "subtitle": "Build three analytical lanes, verify their answers, and supervise the result.", "description": "Create a working free PRACTICE swarm and prepare a scoped live Genie SDK workflow with persistent evidence, failure drills and human approval.", "one_sentence_description": "Create a working free PRACTICE swarm and prepare a scoped live Genie SDK workflow with persistent evidence, failure drills and human approval.", "target_user": "A motivated beginner with permission to work in a private practice folder.", "skill_level": "Intermediate, guided", "estimated_time": "4-6 hours; live account verification additional", "category": "Guided Build", "tags": [ "Databricks", "Genie", "swarm", "parallel analytics", "human approval" ], "prerequisites": [ "Python 3.10+ for offline practice", "Optional: owner-approved Databricks workspace, Unity Catalog sample data, SQL warehouse, Genie and Apps access" ], "required_accounts": [], "required_software": [ "Modern browser", "Python 3.10+ for downloadable drills" ], "required_hardware": [ "A computer with Python 3.10+ and a terminal for practice.", "Internet and owner-approved Databricks workspace access are optional for the separate live track." ], "expected_cost": "$0 required practice; optional cloud use needs approved budget", "final_outcome": "An inspectable three-lane analytical practice swarm with retained reports, failure checks and hash-bound human review; live cloud graduation requires separate observed evidence.", "platforms": [ "windows", "macos", "linux" ], "version": 13, "author": "chatgpt_codex", "source_checked": "2026-10-13", "verification_status": "Windows practice and browser verified; live workspace and macOS/Linux execution unverified", "target_audience": "Non-technical builders and data practitioners" }, "user_outcome": { "before": "Individual analytical questions require manual coordination.", "after": "Create a working free PRACTICE swarm and prepare a scoped live Genie SDK workflow with persistent evidence, failure drills and human approval.", "finished_product": "Inspectable practice runtime, three-space SDK connector, dashboard, failure tests and separate live evidence gate." }, "learning_objectives": [ "Define three scoped analytical work orders.", "Run concurrent lanes and retain state history.", "Reject invalid data and bound retry attempts.", "Bind approval to verified report bytes.", "Prepare authenticated Genie specialist queries without borrowing credentials.", "Distinguish observed practice results from unverified cloud and general autonomy claims." ], "lessons": [ { "lesson_id": "L01", "title": "The swarm transformation", "purpose": "One mission becomes separate analytical work orders, verified results and a human decision. Stages: one bot, multiple bots, AI team, competition, mission swarm, bounded autonomous swarm. A drone analogy means coordinated observation only.", "estimated_time": "15-25 min", "learning_objective": "Python 3.10+ is available; you can explain the difference between practice completion and live graduation.", "lessons_before_this": [], "lessons_after_this": [ "L02" ] }, { "lesson_id": "L02", "title": "Databricks and Genie", "purpose": "Unity Catalog governs data; a Genie Space specializes over approved tables. Databricks Apps hosts your UI; it does not grant automatic access to every table.", "estimated_time": "15-25 min", "learning_objective": "Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred.", "lessons_before_this": [ "L01" ], "lessons_after_this": [ "L03" ] }, { "lesson_id": "L03", "title": "The coordinator", "purpose": "The kit decomposes one named revenue/churn mission through a fixed three-domain recipe. Open-ended AI planning is a separate extension, never inferred from parallel threads.", "estimated_time": "15-25 min", "learning_objective": "Three verified results await human review; cost is 0 and the planner says fixed recipe.", "lessons_before_this": [ "L02" ], "lessons_after_this": [ "L04" ] }, { "lesson_id": "L04", "title": "Persistent task queue", "purpose": "The required SQLite event journal is local memory. Delta SQL is supplied for a separate live implementation; ACID alone does not implement work claiming, locks or exactly-once dispatch.", "estimated_time": "15-25 min", "learning_objective": "The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded.", "lessons_before_this": [ "L03" ], "lessons_after_this": [ "L05" ] }, { "lesson_id": "L05", "title": "The Genie SDK connector", "purpose": "The official SDK waiter polls a conversation until completion; completed text alone does not establish verified query data.", "estimated_time": "15-25 min", "learning_objective": "The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check.", "lessons_before_this": [ "L04" ], "lessons_after_this": [ "L06" ] }, { "lesson_id": "L06", "title": "Parallel analytical lanes", "purpose": "Concurrency is observed overlap, not a count of worker labels. Workspace quotas can constrain live throughput.", "estimated_time": "15-25 min", "learning_objective": "Your timestamps demonstrate the difference between serial and parallel execution.", "lessons_before_this": [ "L05" ], "lessons_after_this": [ "L07" ] }, { "lesson_id": "L07", "title": "Recovery without blind retries", "purpose": "Transient failures can justify bounded retry. Invalid data, missing permissions and uncertain remote completion require diagnosis.", "estimated_time": "15-25 min", "learning_objective": "Mission state is FAILED with no approval; the failed evidence remains available.", "lessons_before_this": [ "L06" ], "lessons_after_this": [ "L08" ] }, { "lesson_id": "L08", "title": "Independent output verification", "purpose": "An executable oracle compares sample totals independently of the specialist answer. It cannot establish the business meaning of arbitrary production data.", "estimated_time": "15-25 min", "learning_objective": "Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle.", "lessons_before_this": [ "L07" ], "lessons_after_this": [ "L09" ] }, { "lesson_id": "L09", "title": "Mission control dashboard", "purpose": "The dashboard shows actual retained reports and local approvals. It runs practice only; no token count or estimated DBU value is fabricated.", "estimated_time": "15-25 min", "learning_objective": "An invalid report has no release action and no invented cloud cost meter.", "lessons_before_this": [ "L08" ], "lessons_after_this": [ "L10" ] }, { "lesson_id": "L10", "title": "Human guardrails and cost", "purpose": "Local call and runtime checks stop new work. Remote queries may outlive client timeouts; hard cloud limits require administrator-enforced controls.", "estimated_time": "15-25 min", "learning_objective": "A receipt binds the unchanged report hash; altered or failed reports are rejected.", "lessons_before_this": [ "L09" ], "lessons_after_this": [ "L11" ] }, { "lesson_id": "L11", "title": "End-to-end forensic walkthrough", "purpose": "Revenue leakage and churn questions can coordinate specialists; causal explanations need more evidence than aggregates.", "estimated_time": "15-25 min", "learning_objective": "Every factual claim maps to a result; unsupported strategic conclusions remain unanswered.", "lessons_before_this": [ "L10" ], "lessons_after_this": [ "L12" ] }, { "lesson_id": "L12", "title": "Deploying and CI/CD", "purpose": "Cloud deployment requires your explicit account authorization and permissions. A service principal needs separate grants; an app definition alone grants none.", "estimated_time": "15-25 min", "learning_objective": "Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate.", "lessons_before_this": [ "L11" ], "lessons_after_this": [ "L13" ] }, { "lesson_id": "L13", "title": "Capstone and Build Bot Better", "purpose": "Graduation requires a new mission and honest scope. Durable memory, owner identity and reproducible failure evidence matter more than a self-awarded grade.", "estimated_time": "15-25 min", "learning_objective": "A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded.", "lessons_before_this": [ "L12" ], "lessons_after_this": [] } ], "steps": [ { "title": "Draw the mission path", "purpose": "One mission becomes separate analytical work orders, verified results and a human decision. Stages: one bot, multiple bots, AI team, competition, mission swarm, bounded autonomous swarm. A drone analogy means coordinated observation only.", "user_action": "Draw human -> coordinator -> queue -> finance/customer/operations specialists -> verifier -> report -> human approval. Mark which components are software and which still need your judgment. The required kit is deterministic PRACTICE, not AI.", "explanation": "Verified results and human release approval. Expected evidence: Your diagram includes verification and approval, and labels practice workers.", "estimated_time": "2-4 min", "action_type": "verify", "commands": { "windows": "", "macos": "", "linux": "" }, "command_type": "shell", "required_inputs": [], "expected_result": "Your diagram includes verification and approval, and labels practice workers.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Draw the mission path", "success_condition": "Your diagram includes verification and approval, and labels practice workers.", "how_to_verify": "Record one recognized evidence group: verification, approval, practice. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "Your diagram includes verification and approval, and labels practice workers.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Your diagram includes verification and approval, and labels practice workers.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L01-S01-CHECK", "tokens_any": [ [ "verification", "approval", "practice" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Draw the mission path, what exact command or inspection failed, and how did its output differ from Your diagram includes verification and approval, and labels practice workers.?", "fix": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Your diagram includes verification and approval, and labels practice workers. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "Which parts of the diagram need explicit gates?", "answer": "Verified results and human release approval. Your diagram includes verification and approval, and labels practice workers." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "Which parts of the diagram need explicit gates?", "choices": [ "Verified results and human release approval.", "Only the final formatting." ], "correct_answer": 0, "explanation": "Verified results and human release approval." }, "coach": { "intro": "Next: Draw the mission path", "demo": "Draw human -> coordinator -> queue -> finance/customer/operations specialists -> verifier -> report -> human approval. Mark which components are software and which still need your judgment. The required kit is deterministic PRACTICE, not AI.", "handoff": "Try the action and save what actually happens.", "success": "Your diagram includes verification and approval, and labels practice workers.", "failure": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Your diagram includes verification and approval, and labels practice workers. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Your diagram includes verification and approval, and labels practice workers. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Draw the mission path", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "One mission becomes separate analytical work orders, verified results and a human decision. Stages: one bot, multiple bots, AI team, competition, mission swarm, bounded autonomous swarm. A drone analogy means coordinated observation only. Draw human -> coordinator -> queue -> finance/customer/operations specialists -> verifier -> report -> human approval. Mark which components are software and which still need your judgment. The required kit is deterministic PRACTICE, not AI. Check: Your diagram includes verification and approval, and labels practice workers.", "on_screen_text": [ "Draw the mission path", "Your diagram includes verification and approval, and labels practice workers." ], "screen_recording_instructions": [ "Draw human -> coordinator -> queue -> finance/customer/operations specialists -> verifier -> report -> human approval. Mark which components are software and which still need your judgment. The required kit is deterministic PRACTICE, not AI.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Your diagram includes verification and approval, and labels practice workers." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "One mission becomes separate analytical work orders, verified results and a human decision. Stages: one bot, multiple bots, AI team, competition, mission swarm, bounded autonomous swarm. A drone analogy means coordinated observation only.", "Draw human -> coordinator -> queue -> finance/customer/operations specialists -> verifier -> report -> human approval. Mark which components are software and which still need your judgment. The required kit is deterministic PRACTICE, not AI.", "Your diagram includes verification and approval, and labels practice workers." ] }, "step_id": "L01-S01", "lesson_id": "L01" }, { "title": "Choose a completion track", "purpose": "One mission becomes separate analytical work orders, verified results and a human decision. Stages: one bot, multiple bots, AI team, competition, mission swarm, bounded autonomous swarm. A drone analogy means coordinated observation only.", "user_action": "Download Resources and extract into a new bbb-genie-practice folder outside shared or profile folders. Read README.md. Choose free offline practice or a separately authorized Databricks workspace. Live sign-in and spend are never required for practice.", "explanation": "No; they are separate evidence tracks. Expected evidence: Python 3.10+ is available; you can explain the difference between practice completion and live graduation.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python --version", "macos": "python3 --version", "linux": "python3 --version" }, "command_type": "shell", "required_inputs": [], "expected_result": "Python 3.10+ is available; you can explain the difference between practice completion and live graduation.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Choose a completion track", "success_condition": "Python 3.10+ is available; you can explain the difference between practice completion and live graduation.", "how_to_verify": "Record one recognized evidence group: Python 3.. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "Python 3.10+ is available; you can explain the difference between practice completion and live graduation.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Python 3.10+ is available; you can explain the difference between practice completion and live graduation.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L01-S02-CHECK", "tokens_any": [ [ "Python 3." ] ], "deferred_min_reason_chars": 24, "python_min": [ 3, 10 ] }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Choose a completion track, what exact command or inspection failed, and how did its output differ from Python 3.10+ is available; you can explain the difference between practice completion and live graduation.?", "fix": "For this step run python --version from the extracted kit folder. Compare against: Python 3.10+ is available; you can explain the difference between practice completion and live graduation. If python is not found on Windows, try py -3 --version; Python 3.10 or newer is required. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "Can offline practice earn live cloud graduation?", "answer": "No; they are separate evidence tracks. Python 3.10+ is available; you can explain the difference between practice completion and live graduation." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "Can offline practice earn live cloud graduation?", "choices": [ "Yes; local practice proves cloud access.", "No; they are separate evidence tracks." ], "correct_answer": 1, "explanation": "No; they are separate evidence tracks." }, "coach": { "intro": "Next: Choose a completion track", "demo": "Download Resources and extract into a new bbb-genie-practice folder outside shared or profile folders. Read README.md. Choose free offline practice or a separately authorized Databricks workspace. Live sign-in and spend are never required for practice.", "handoff": "Try the action and save what actually happens.", "success": "Python 3.10+ is available; you can explain the difference between practice completion and live graduation.", "failure": "For this step run python --version from the extracted kit folder. Compare against: Python 3.10+ is available; you can explain the difference between practice completion and live graduation. If python is not found on Windows, try py -3 --version; Python 3.10 or newer is required. ", "help": "For this step run python --version from the extracted kit folder. Compare against: Python 3.10+ is available; you can explain the difference between practice completion and live graduation. If python is not found on Windows, try py -3 --version; Python 3.10 or newer is required. " }, "video": { "purpose": "Choose a completion track", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "One mission becomes separate analytical work orders, verified results and a human decision. Stages: one bot, multiple bots, AI team, competition, mission swarm, bounded autonomous swarm. A drone analogy means coordinated observation only. Download Resources and extract into a new bbb-genie-practice folder outside shared or profile folders. Read README.md. Choose free offline practice or a separately authorized Databricks workspace. Live sign-in and spend are never required for practice. Check: Python 3.10+ is available; you can explain the difference between practice completion and live graduation.", "on_screen_text": [ "Choose a completion track", "Python 3.10+ is available; you can explain the difference between practice completion and live graduation." ], "screen_recording_instructions": [ "Download Resources and extract into a new bbb-genie-practice folder outside shared or profile folders. Read README.md. Choose free offline practice or a separately authorized Databricks workspace. Live sign-in and spend are never required for practice.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Python 3.10+ is available; you can explain the difference between practice completion and live graduation." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "One mission becomes separate analytical work orders, verified results and a human decision. Stages: one bot, multiple bots, AI team, competition, mission swarm, bounded autonomous swarm. A drone analogy means coordinated observation only.", "Download Resources and extract into a new bbb-genie-practice folder outside shared or profile folders. Read README.md. Choose free offline practice or a separately authorized Databricks workspace. Live sign-in and spend are never required for practice.", "Python 3.10+ is available; you can explain the difference between practice completion and live graduation." ] }, "step_id": "L01-S02", "lesson_id": "L01" }, { "title": "Check workspace prerequisites", "purpose": "Unity Catalog governs data; a Genie Space specializes over approved tables. Databricks Apps hosts your UI; it does not grant automatic access to every table.", "user_action": "For live work ask the workspace owner to confirm Unity Catalog, an accessible SQL warehouse, Genie availability, Databricks Apps availability and an approved spend plan. Record cloud and region. If absent, continue practice and mark live steps deferred. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "explanation": "The workspace owner under your own identity. Expected evidence: An access checklist names each available or deferred feature without recording secrets.", "estimated_time": "2-4 min", "action_type": "verify", "commands": { "windows": "", "macos": "", "linux": "" }, "command_type": "shell", "required_inputs": [], "expected_result": "An access checklist names each available or deferred feature without recording secrets.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Check workspace prerequisites", "success_condition": "An access checklist names each available or deferred feature without recording secrets.", "how_to_verify": "Record one recognized evidence group: Unity Catalog, warehouse, Genie OR DEFERRED, reason. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "An access checklist names each available or deferred feature without recording secrets.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: An access checklist names each available or deferred feature without recording secrets.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L02-S01-CHECK", "tokens_any": [ [ "Unity Catalog", "warehouse", "Genie" ], [ "DEFERRED", "reason" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Check workspace prerequisites, what exact command or inspection failed, and how did its output differ from An access checklist names each available or deferred feature without recording secrets.?", "fix": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: An access checklist names each available or deferred feature without recording secrets. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "Who confirms live feature access and spending?", "answer": "The workspace owner under your own identity. An access checklist names each available or deferred feature without recording secrets." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "Who confirms live feature access and spending?", "choices": [ "The workspace owner under your own identity.", "A role name in the diagram." ], "correct_answer": 0, "explanation": "The workspace owner under your own identity." }, "coach": { "intro": "Next: Check workspace prerequisites", "demo": "For live work ask the workspace owner to confirm Unity Catalog, an accessible SQL warehouse, Genie availability, Databricks Apps availability and an approved spend plan. Record cloud and region. If absent, continue practice and mark live steps deferred. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "handoff": "Try the action and save what actually happens.", "success": "An access checklist names each available or deferred feature without recording secrets.", "failure": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: An access checklist names each available or deferred feature without recording secrets. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: An access checklist names each available or deferred feature without recording secrets. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Check workspace prerequisites", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Unity Catalog governs data; a Genie Space specializes over approved tables. Databricks Apps hosts your UI; it does not grant automatic access to every table. For live work ask the workspace owner to confirm Unity Catalog, an accessible SQL warehouse, Genie availability, Databricks Apps availability and an approved spend plan. Record cloud and region. If absent, continue practice and mark live steps deferred. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution. Check: An access checklist names each available or deferred feature without recording secrets.", "on_screen_text": [ "Check workspace prerequisites", "An access checklist names each available or deferred feature without recording secrets." ], "screen_recording_instructions": [ "For live work ask the workspace owner to confirm Unity Catalog, an accessible SQL warehouse, Genie availability, Databricks Apps availability and an approved spend plan. Record cloud and region. If absent, continue practice and mark live steps deferred. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "An access checklist names each available or deferred feature without recording secrets." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Unity Catalog governs data; a Genie Space specializes over approved tables. Databricks Apps hosts your UI; it does not grant automatic access to every table.", "For live work ask the workspace owner to confirm Unity Catalog, an accessible SQL warehouse, Genie availability, Databricks Apps availability and an approved spend plan. Record cloud and region. If absent, continue practice and mark live steps deferred. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "An access checklist names each available or deferred feature without recording secrets." ] }, "step_id": "L02-S01", "lesson_id": "L02" }, { "title": "Create three scoped specialists", "purpose": "Unity Catalog governs data; a Genie Space specializes over approved tables. Databricks Apps hosts your UI; it does not grant automatic access to every table.", "user_action": "In Genie create an agent/space for finance, customers and operations. Choose your sample table, SQL warehouse and descriptions from workspace-guide.md. Add the exact metric definitions. Run each question directly before wiring orchestration. Do not add production tables. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "explanation": "Its agreed metric on the approved sample dataset. Expected evidence: Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred.", "estimated_time": "2-4 min", "action_type": "verify", "commands": { "windows": "", "macos": "", "linux": "" }, "command_type": "shell", "required_inputs": [], "expected_result": "Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Create three scoped specialists", "success_condition": "Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred.", "how_to_verify": "Record one recognized evidence group: revenue, 3000, churned, 3, late_orders, 4 OR DEFERRED, reason. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L02-S02-CHECK", "tokens_any": [ [ "revenue", "3000", "churned", "3", "late_orders", "4" ], [ "DEFERRED", "reason" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Create three scoped specialists, what exact command or inspection failed, and how did its output differ from Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred.?", "fix": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What should a specialist query establish?", "answer": "Its agreed metric on the approved sample dataset. Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What should a specialist query establish?", "choices": [ "Access to every production table.", "Its agreed metric on the approved sample dataset." ], "correct_answer": 1, "explanation": "Its agreed metric on the approved sample dataset." }, "coach": { "intro": "Next: Create three scoped specialists", "demo": "In Genie create an agent/space for finance, customers and operations. Choose your sample table, SQL warehouse and descriptions from workspace-guide.md. Add the exact metric definitions. Run each question directly before wiring orchestration. Do not add production tables. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "handoff": "Try the action and save what actually happens.", "success": "Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred.", "failure": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Create three scoped specialists", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Unity Catalog governs data; a Genie Space specializes over approved tables. Databricks Apps hosts your UI; it does not grant automatic access to every table. In Genie create an agent/space for finance, customers and operations. Choose your sample table, SQL warehouse and descriptions from workspace-guide.md. Add the exact metric definitions. Run each question directly before wiring orchestration. Do not add production tables. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution. Check: Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred.", "on_screen_text": [ "Create three scoped specialists", "Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred." ], "screen_recording_instructions": [ "In Genie create an agent/space for finance, customers and operations. Choose your sample table, SQL warehouse and descriptions from workspace-guide.md. Add the exact metric definitions. Run each question directly before wiring orchestration. Do not add production tables. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Unity Catalog governs data; a Genie Space specializes over approved tables. Databricks Apps hosts your UI; it does not grant automatic access to every table.", "In Genie create an agent/space for finance, customers and operations. Choose your sample table, SQL warehouse and descriptions from workspace-guide.md. Add the exact metric definitions. Run each question directly before wiring orchestration. Do not add production tables. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "Each specialist returns its single expected sample total: revenue 3000, churned 3, late_orders 4; or the live check is explicitly deferred." ] }, "step_id": "L02-S02", "lesson_id": "L02" }, { "title": "Inspect the work orders", "purpose": "The kit decomposes one named revenue/churn mission through a fixed three-domain recipe. Open-ended AI planning is a separate extension, never inferred from parallel threads.", "user_action": "Open swarm.py and read QUESTIONS and EXPECTED. Match each work order to one domain. Copy a mission contract into evidence.md: dataset, period, three questions, runtime, call count, acceptance conditions and approval owner.", "explanation": "A fixed recipe for three analytical lanes. Expected evidence: Three domain questions share the same sample period and population.", "estimated_time": "2-4 min", "action_type": "verify", "commands": { "windows": "", "macos": "", "linux": "" }, "command_type": "shell", "required_inputs": [], "expected_result": "Three domain questions share the same sample period and population.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Inspect the work orders", "success_condition": "Three domain questions share the same sample period and population.", "how_to_verify": "Record one recognized evidence group: finance, customers, operations. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "Three domain questions share the same sample period and population.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Three domain questions share the same sample period and population.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L03-S01-CHECK", "tokens_any": [ [ "finance", "customers", "operations" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Inspect the work orders, what exact command or inspection failed, and how did its output differ from Three domain questions share the same sample period and population.?", "fix": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Three domain questions share the same sample period and population. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What kind of planner ships in this kit?", "answer": "A fixed recipe for three analytical lanes. Three domain questions share the same sample period and population." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What kind of planner ships in this kit?", "choices": [ "A fixed recipe for three analytical lanes.", "An arbitrary general mission planner." ], "correct_answer": 0, "explanation": "A fixed recipe for three analytical lanes." }, "coach": { "intro": "Next: Inspect the work orders", "demo": "Open swarm.py and read QUESTIONS and EXPECTED. Match each work order to one domain. Copy a mission contract into evidence.md: dataset, period, three questions, runtime, call count, acceptance conditions and approval owner.", "handoff": "Try the action and save what actually happens.", "success": "Three domain questions share the same sample period and population.", "failure": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Three domain questions share the same sample period and population. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Three domain questions share the same sample period and population. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Inspect the work orders", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "The kit decomposes one named revenue/churn mission through a fixed three-domain recipe. Open-ended AI planning is a separate extension, never inferred from parallel threads. Open swarm.py and read QUESTIONS and EXPECTED. Match each work order to one domain. Copy a mission contract into evidence.md: dataset, period, three questions, runtime, call count, acceptance conditions and approval owner. Check: Three domain questions share the same sample period and population.", "on_screen_text": [ "Inspect the work orders", "Three domain questions share the same sample period and population." ], "screen_recording_instructions": [ "Open swarm.py and read QUESTIONS and EXPECTED. Match each work order to one domain. Copy a mission contract into evidence.md: dataset, period, three questions, runtime, call count, acceptance conditions and approval owner.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Three domain questions share the same sample period and population." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "The kit decomposes one named revenue/churn mission through a fixed three-domain recipe. Open-ended AI planning is a separate extension, never inferred from parallel threads.", "Open swarm.py and read QUESTIONS and EXPECTED. Match each work order to one domain. Copy a mission contract into evidence.md: dataset, period, three questions, runtime, call count, acceptance conditions and approval owner.", "Three domain questions share the same sample period and population." ] }, "step_id": "L03-S01", "lesson_id": "L03" }, { "title": "Run your first mission", "purpose": "The kit decomposes one named revenue/churn mission through a fixed three-domain recipe. Open-ended AI planning is a separate extension, never inferred from parallel threads.", "user_action": "In the extracted folder run the command. Open the new JSON under runs. Identify the PRACTICE label, three results, call count, cost and AWAITING_HUMAN state. The report is a factual metric summary, not an investment recommendation.", "explanation": "After verified bytes receive human approval. Expected evidence: Three verified results await human review; cost is 0 and the planner says fixed recipe.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python swarm.py", "macos": "python3 swarm.py", "linux": "python3 swarm.py" }, "command_type": "shell", "required_inputs": [], "expected_result": "Three verified results await human review; cost is 0 and the planner says fixed recipe.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Run your first mission", "success_condition": "Three verified results await human review; cost is 0 and the planner says fixed recipe.", "how_to_verify": "Record one recognized evidence group: PRACTICE, AWAITING_HUMAN, REPORT_SHA256. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "Three verified results await human review; cost is 0 and the planner says fixed recipe.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Three verified results await human review; cost is 0 and the planner says fixed recipe.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L03-S02-CHECK", "tokens_any": [ [ "PRACTICE", "AWAITING_HUMAN", "REPORT_SHA256" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Run your first mission, what exact command or inspection failed, and how did its output differ from Three verified results await human review; cost is 0 and the planner says fixed recipe.?", "fix": "For this step run python swarm.py from the extracted kit folder. Compare against: Three verified results await human review; cost is 0 and the planner says fixed recipe. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "When is the first report released?", "answer": "After verified bytes receive human approval. Three verified results await human review; cost is 0 and the planner says fixed recipe." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "When is the first report released?", "choices": [ "When the workers have returned strings.", "After verified bytes receive human approval." ], "correct_answer": 1, "explanation": "After verified bytes receive human approval." }, "coach": { "intro": "Next: Run your first mission", "demo": "In the extracted folder run the command. Open the new JSON under runs. Identify the PRACTICE label, three results, call count, cost and AWAITING_HUMAN state. The report is a factual metric summary, not an investment recommendation.", "handoff": "Try the action and save what actually happens.", "success": "Three verified results await human review; cost is 0 and the planner says fixed recipe.", "failure": "For this step run python swarm.py from the extracted kit folder. Compare against: Three verified results await human review; cost is 0 and the planner says fixed recipe. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "For this step run python swarm.py from the extracted kit folder. Compare against: Three verified results await human review; cost is 0 and the planner says fixed recipe. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Run your first mission", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "The kit decomposes one named revenue/churn mission through a fixed three-domain recipe. Open-ended AI planning is a separate extension, never inferred from parallel threads. In the extracted folder run the command. Open the new JSON under runs. Identify the PRACTICE label, three results, call count, cost and AWAITING_HUMAN state. The report is a factual metric summary, not an investment recommendation. Check: Three verified results await human review; cost is 0 and the planner says fixed recipe.", "on_screen_text": [ "Run your first mission", "Three verified results await human review; cost is 0 and the planner says fixed recipe." ], "screen_recording_instructions": [ "In the extracted folder run the command. Open the new JSON under runs. Identify the PRACTICE label, three results, call count, cost and AWAITING_HUMAN state. The report is a factual metric summary, not an investment recommendation.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Three verified results await human review; cost is 0 and the planner says fixed recipe." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "The kit decomposes one named revenue/churn mission through a fixed three-domain recipe. Open-ended AI planning is a separate extension, never inferred from parallel threads.", "In the extracted folder run the command. Open the new JSON under runs. Identify the PRACTICE label, three results, call count, cost and AWAITING_HUMAN state. The report is a factual metric summary, not an investment recommendation.", "Three verified results await human review; cost is 0 and the planner says fixed recipe." ] }, "step_id": "L03-S02", "lesson_id": "L03" }, { "title": "Inspect task transitions", "purpose": "The required SQLite event journal is local memory. Delta SQL is supplied for a separate live implementation; ACID alone does not implement work claiming, locks or exactly-once dispatch.", "user_action": "Open memory-guide.md and use the command to list events. For every lane locate QUEUED, ASSIGNED, RUNNING, VERIFYING and COMPLETED. A failed lane must never be silently marked completed.", "explanation": "Its timestamped run events. Expected evidence: Every successful lane has timestamped transitions associated with a unique run ID.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python inspect_memory.py", "macos": "python3 inspect_memory.py", "linux": "python3 inspect_memory.py" }, "command_type": "shell", "required_inputs": [], "expected_result": "Every successful lane has timestamped transitions associated with a unique run ID.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Inspect task transitions", "success_condition": "Every successful lane has timestamped transitions associated with a unique run ID.", "how_to_verify": "Record one recognized evidence group: QUEUED, RUNNING, VERIFYING, COMPLETED. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "Every successful lane has timestamped transitions associated with a unique run ID.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Every successful lane has timestamped transitions associated with a unique run ID.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L04-S01-CHECK", "tokens_any": [ [ "QUEUED", "RUNNING", "VERIFYING", "COMPLETED" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Inspect task transitions, what exact command or inspection failed, and how did its output differ from Every successful lane has timestamped transitions associated with a unique run ID.?", "fix": "For this step run python inspect_memory.py from the extracted kit folder. Compare against: Every successful lane has timestamped transitions associated with a unique run ID. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What proves a lane progressed?", "answer": "Its timestamped run events. Every successful lane has timestamped transitions associated with a unique run ID." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What proves a lane progressed?", "choices": [ "Its timestamped run events.", "A permanently green diagram." ], "correct_answer": 0, "explanation": "Its timestamped run events." }, "coach": { "intro": "Next: Inspect task transitions", "demo": "Open memory-guide.md and use the command to list events. For every lane locate QUEUED, ASSIGNED, RUNNING, VERIFYING and COMPLETED. A failed lane must never be silently marked completed.", "handoff": "Try the action and save what actually happens.", "success": "Every successful lane has timestamped transitions associated with a unique run ID.", "failure": "For this step run python inspect_memory.py from the extracted kit folder. Compare against: Every successful lane has timestamped transitions associated with a unique run ID. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "For this step run python inspect_memory.py from the extracted kit folder. Compare against: Every successful lane has timestamped transitions associated with a unique run ID. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Inspect task transitions", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "The required SQLite event journal is local memory. Delta SQL is supplied for a separate live implementation; ACID alone does not implement work claiming, locks or exactly-once dispatch. Open memory-guide.md and use the command to list events. For every lane locate QUEUED, ASSIGNED, RUNNING, VERIFYING and COMPLETED. A failed lane must never be silently marked completed. Check: Every successful lane has timestamped transitions associated with a unique run ID.", "on_screen_text": [ "Inspect task transitions", "Every successful lane has timestamped transitions associated with a unique run ID." ], "screen_recording_instructions": [ "Open memory-guide.md and use the command to list events. For every lane locate QUEUED, ASSIGNED, RUNNING, VERIFYING and COMPLETED. A failed lane must never be silently marked completed.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Every successful lane has timestamped transitions associated with a unique run ID." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "The required SQLite event journal is local memory. Delta SQL is supplied for a separate live implementation; ACID alone does not implement work claiming, locks or exactly-once dispatch.", "Open memory-guide.md and use the command to list events. For every lane locate QUEUED, ASSIGNED, RUNNING, VERIFYING and COMPLETED. A failed lane must never be silently marked completed.", "Every successful lane has timestamped transitions associated with a unique run ID." ] }, "step_id": "L04-S01", "lesson_id": "L04" }, { "title": "Prepare Delta memory deliberately", "purpose": "The required SQLite event journal is local memory. Delta SQL is supplied for a separate live implementation; ACID alone does not implement work claiming, locks or exactly-once dispatch.", "user_action": "Read queue.sql and replace YOUR_CATALOG and YOUR_SCHEMA with an owner-approved sandbox. Run only in an authorized Databricks SQL editor. This creates new swarm event tables; it does not connect the local runtime. Record the remaining adapter work in evidence.md. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "explanation": "No; a claim/lease adapter is still required. Expected evidence: The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded.", "estimated_time": "2-4 min", "action_type": "verify", "commands": { "windows": "", "macos": "", "linux": "" }, "command_type": "shell", "required_inputs": [], "expected_result": "The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Prepare Delta memory deliberately", "success_condition": "The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded.", "how_to_verify": "Record one recognized evidence group: Delta, claim, lease OR DEFERRED, reason. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L04-S02-CHECK", "tokens_any": [ [ "Delta", "claim", "lease" ], [ "DEFERRED", "reason" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Prepare Delta memory deliberately, what exact command or inspection failed, and how did its output differ from The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded.?", "fix": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "Do Delta table templates connect the runtime?", "answer": "No; a claim/lease adapter is still required. The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "Do Delta table templates connect the runtime?", "choices": [ "Yes; SQL tables alone dispatch workers.", "No; a claim/lease adapter is still required." ], "correct_answer": 1, "explanation": "No; a claim/lease adapter is still required." }, "coach": { "intro": "Next: Prepare Delta memory deliberately", "demo": "Read queue.sql and replace YOUR_CATALOG and YOUR_SCHEMA with an owner-approved sandbox. Run only in an authorized Databricks SQL editor. This creates new swarm event tables; it does not connect the local runtime. Record the remaining adapter work in evidence.md. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "handoff": "Try the action and save what actually happens.", "success": "The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded.", "failure": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Prepare Delta memory deliberately", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "The required SQLite event journal is local memory. Delta SQL is supplied for a separate live implementation; ACID alone does not implement work claiming, locks or exactly-once dispatch. Read queue.sql and replace YOUR_CATALOG and YOUR_SCHEMA with an owner-approved sandbox. Run only in an authorized Databricks SQL editor. This creates new swarm event tables; it does not connect the local runtime. Record the remaining adapter work in evidence.md. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution. Check: The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded.", "on_screen_text": [ "Prepare Delta memory deliberately", "The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded." ], "screen_recording_instructions": [ "Read queue.sql and replace YOUR_CATALOG and YOUR_SCHEMA with an owner-approved sandbox. Run only in an authorized Databricks SQL editor. This creates new swarm event tables; it does not connect the local runtime. Record the remaining adapter work in evidence.md. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "The required SQLite event journal is local memory. Delta SQL is supplied for a separate live implementation; ACID alone does not implement work claiming, locks or exactly-once dispatch.", "Read queue.sql and replace YOUR_CATALOG and YOUR_SCHEMA with an owner-approved sandbox. Run only in an authorized Databricks SQL editor. This creates new swarm event tables; it does not connect the local runtime. Record the remaining adapter work in evidence.md. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "The two Delta tables exist in the sandbox, or live setup is deferred; the missing distributed claim/lease adapter is recorded." ] }, "step_id": "L04-S02", "lesson_id": "L04" }, { "title": "Authenticate under your identity", "purpose": "The official SDK waiter polls a conversation until completion; completed text alone does not establish verified query data.", "user_action": "Read workspace-guide.md. Create a virtual environment, install requirements and authenticate the Databricks CLI using OAuth under your own identity. Use databricks auth describe to confirm the host and identity locally. Never paste tokens into course evidence. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "explanation": "A redacted identity/configuration observation, never a token. Expected evidence: The dependency installation succeeds; live identity is confirmed locally or explicitly deferred.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python -m pip install -r requirements.txt", "macos": "python3 -m pip install -r requirements.txt", "linux": "python3 -m pip install -r requirements.txt" }, "command_type": "shell", "required_inputs": [], "expected_result": "The dependency installation succeeds; live identity is confirmed locally or explicitly deferred.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Authenticate under your identity", "success_condition": "The dependency installation succeeds; live identity is confirmed locally or explicitly deferred.", "how_to_verify": "Record one recognized evidence group: Successfully installed OR Requirement already satisfied OR DEFERRED, reason. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "The dependency installation succeeds; live identity is confirmed locally or explicitly deferred.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: The dependency installation succeeds; live identity is confirmed locally or explicitly deferred.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L05-S01-CHECK", "tokens_any": [ [ "Successfully installed" ], [ "Requirement already satisfied" ], [ "DEFERRED", "reason" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Authenticate under your identity, what exact command or inspection failed, and how did its output differ from The dependency installation succeeds; live identity is confirmed locally or explicitly deferred.?", "fix": "For this step run python -m pip install -r requirements.txt from the extracted kit folder. Compare against: The dependency installation succeeds; live identity is confirmed locally or explicitly deferred. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What may you paste as authentication evidence?", "answer": "A redacted identity/configuration observation, never a token. The dependency installation succeeds; live identity is confirmed locally or explicitly deferred." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What may you paste as authentication evidence?", "choices": [ "A redacted identity/configuration observation, never a token.", "Your full token or password." ], "correct_answer": 0, "explanation": "A redacted identity/configuration observation, never a token." }, "coach": { "intro": "Next: Authenticate under your identity", "demo": "Read workspace-guide.md. Create a virtual environment, install requirements and authenticate the Databricks CLI using OAuth under your own identity. Use databricks auth describe to confirm the host and identity locally. Never paste tokens into course evidence. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "handoff": "Try the action and save what actually happens.", "success": "The dependency installation succeeds; live identity is confirmed locally or explicitly deferred.", "failure": "For this step run python -m pip install -r requirements.txt from the extracted kit folder. Compare against: The dependency installation succeeds; live identity is confirmed locally or explicitly deferred. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "For this step run python -m pip install -r requirements.txt from the extracted kit folder. Compare against: The dependency installation succeeds; live identity is confirmed locally or explicitly deferred. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Authenticate under your identity", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "The official SDK waiter polls a conversation until completion; completed text alone does not establish verified query data. Read workspace-guide.md. Create a virtual environment, install requirements and authenticate the Databricks CLI using OAuth under your own identity. Use databricks auth describe to confirm the host and identity locally. Never paste tokens into course evidence. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution. Check: The dependency installation succeeds; live identity is confirmed locally or explicitly deferred.", "on_screen_text": [ "Authenticate under your identity", "The dependency installation succeeds; live identity is confirmed locally or explicitly deferred." ], "screen_recording_instructions": [ "Read workspace-guide.md. Create a virtual environment, install requirements and authenticate the Databricks CLI using OAuth under your own identity. Use databricks auth describe to confirm the host and identity locally. Never paste tokens into course evidence. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "The dependency installation succeeds; live identity is confirmed locally or explicitly deferred." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "The official SDK waiter polls a conversation until completion; completed text alone does not establish verified query data.", "Read workspace-guide.md. Create a virtual environment, install requirements and authenticate the Databricks CLI using OAuth under your own identity. Use databricks auth describe to confirm the host and identity locally. Never paste tokens into course evidence. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "The dependency installation succeeds; live identity is confirmed locally or explicitly deferred." ] }, "step_id": "L05-S01", "lesson_id": "L05" }, { "title": "Connect and query the spaces", "purpose": "The official SDK waiter polls a conversation until completion; completed text alone does not establish verified query data.", "user_action": "Copy config.example.json to config.json and enter the three real space IDs. Inspect Live.query in swarm.py: start_conversation_and_wait, query attachment, attachment results and provenance IDs. Run only after the owner approves cloud use. A timeout does not cancel remote SQL. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "explanation": "Actual conversation and query provenance from your authorized workspace. Expected evidence: The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python swarm.py --mode live --deadline 120", "macos": "python3 swarm.py --mode live --deadline 120", "linux": "python3 swarm.py --mode live --deadline 120" }, "command_type": "shell", "required_inputs": [], "expected_result": "The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Connect and query the spaces", "success_condition": "The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check.", "how_to_verify": "Record one recognized evidence group: LIVE, conversation_id, attachment_id OR DEFERRED, reason. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L05-S02-CHECK", "tokens_any": [ [ "LIVE", "conversation_id", "attachment_id" ], [ "DEFERRED", "reason" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Connect and query the spaces, what exact command or inspection failed, and how did its output differ from The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check.?", "fix": "For this step run python swarm.py --mode live --deadline 120 from the extracted kit folder. Compare against: The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check. If config.json is missing, copy config.example.json and supply approved space IDs. Authentication or permission failure needs your own identity and workspace-owner grants; do not substitute PRACTICE. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What distinguishes live query evidence?", "answer": "Actual conversation and query provenance from your authorized workspace. The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What distinguishes live query evidence?", "choices": [ "A practice label changed to LIVE.", "Actual conversation and query provenance from your authorized workspace." ], "correct_answer": 1, "explanation": "Actual conversation and query provenance from your authorized workspace." }, "coach": { "intro": "Next: Connect and query the spaces", "demo": "Copy config.example.json to config.json and enter the three real space IDs. Inspect Live.query in swarm.py: start_conversation_and_wait, query attachment, attachment results and provenance IDs. Run only after the owner approves cloud use. A timeout does not cancel remote SQL. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "handoff": "Try the action and save what actually happens.", "success": "The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check.", "failure": "For this step run python swarm.py --mode live --deadline 120 from the extracted kit folder. Compare against: The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check. If config.json is missing, copy config.example.json and supply approved space IDs. Authentication or permission failure needs your own identity and workspace-owner grants; do not substitute PRACTICE. ", "help": "For this step run python swarm.py --mode live --deadline 120 from the extracted kit folder. Compare against: The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check. If config.json is missing, copy config.example.json and supply approved space IDs. Authentication or permission failure needs your own identity and workspace-owner grants; do not substitute PRACTICE. " }, "video": { "purpose": "Connect and query the spaces", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "The official SDK waiter polls a conversation until completion; completed text alone does not establish verified query data. Copy config.example.json to config.json and enter the three real space IDs. Inspect Live.query in swarm.py: start_conversation_and_wait, query attachment, attachment results and provenance IDs. Run only after the owner approves cloud use. A timeout does not cancel remote SQL. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution. Check: The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check.", "on_screen_text": [ "Connect and query the spaces", "The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check." ], "screen_recording_instructions": [ "Copy config.example.json to config.json and enter the three real space IDs. Inspect Live.query in swarm.py: start_conversation_and_wait, query attachment, attachment results and provenance IDs. Run only after the owner approves cloud use. A timeout does not cancel remote SQL. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "The official SDK waiter polls a conversation until completion; completed text alone does not establish verified query data.", "Copy config.example.json to config.json and enter the three real space IDs. Inspect Live.query in swarm.py: start_conversation_and_wait, query attachment, attachment results and provenance IDs. Run only after the owner approves cloud use. A timeout does not cancel remote SQL. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "The report says LIVE and includes conversation/message/attachment IDs and verified totals, or a diagnosed failed/deferred live check." ] }, "step_id": "L05-S02", "lesson_id": "L05" }, { "title": "Prove overlap", "purpose": "Concurrency is observed overlap, not a count of worker labels. Workspace quotas can constrain live throughput.", "user_action": "Run the acceptance command. It compares the latest worker start with the earliest completion. Save the actual PASS output. The test uses temporary practice folders and retains their contents.", "explanation": "Measured start/end intervals. Expected evidence: The overlap assertion passes and records actual simultaneous PRACTICE lanes.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python test_swarm.py", "macos": "python3 test_swarm.py", "linux": "python3 test_swarm.py" }, "command_type": "shell", "required_inputs": [], "expected_result": "The overlap assertion passes and records actual simultaneous PRACTICE lanes.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Prove overlap", "success_condition": "The overlap assertion passes and records actual simultaneous PRACTICE lanes.", "how_to_verify": "Record one recognized evidence group: PASS: parallel overlap. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "The overlap assertion passes and records actual simultaneous PRACTICE lanes.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: The overlap assertion passes and records actual simultaneous PRACTICE lanes.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L06-S01-CHECK", "tokens_any": [ [ "PASS: parallel overlap" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Prove overlap, what exact command or inspection failed, and how did its output differ from The overlap assertion passes and records actual simultaneous PRACTICE lanes.?", "fix": "For this step run python test_swarm.py from the extracted kit folder. Compare against: The overlap assertion passes and records actual simultaneous PRACTICE lanes. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What proves parallel lanes overlap?", "answer": "Measured start/end intervals. The overlap assertion passes and records actual simultaneous PRACTICE lanes." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What proves parallel lanes overlap?", "choices": [ "Measured start/end intervals.", "Three names in the role list." ], "correct_answer": 0, "explanation": "Measured start/end intervals." }, "coach": { "intro": "Next: Prove overlap", "demo": "Run the acceptance command. It compares the latest worker start with the earliest completion. Save the actual PASS output. The test uses temporary practice folders and retains their contents.", "handoff": "Try the action and save what actually happens.", "success": "The overlap assertion passes and records actual simultaneous PRACTICE lanes.", "failure": "For this step run python test_swarm.py from the extracted kit folder. Compare against: The overlap assertion passes and records actual simultaneous PRACTICE lanes. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "For this step run python test_swarm.py from the extracted kit folder. Compare against: The overlap assertion passes and records actual simultaneous PRACTICE lanes. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Prove overlap", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Concurrency is observed overlap, not a count of worker labels. Workspace quotas can constrain live throughput. Run the acceptance command. It compares the latest worker start with the earliest completion. Save the actual PASS output. The test uses temporary practice folders and retains their contents. Check: The overlap assertion passes and records actual simultaneous PRACTICE lanes.", "on_screen_text": [ "Prove overlap", "The overlap assertion passes and records actual simultaneous PRACTICE lanes." ], "screen_recording_instructions": [ "Run the acceptance command. It compares the latest worker start with the earliest completion. Save the actual PASS output. The test uses temporary practice folders and retains their contents.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "The overlap assertion passes and records actual simultaneous PRACTICE lanes." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Concurrency is observed overlap, not a count of worker labels. Workspace quotas can constrain live throughput.", "Run the acceptance command. It compares the latest worker start with the earliest completion. Save the actual PASS output. The test uses temporary practice folders and retains their contents.", "The overlap assertion passes and records actual simultaneous PRACTICE lanes." ] }, "step_id": "L06-S01", "lesson_id": "L06" }, { "title": "Set a quota-aware boundary", "purpose": "Concurrency is observed overlap, not a count of worker labels. Workspace quotas can constrain live throughput.", "user_action": "Read the ThreadPoolExecutor max_workers=3 setting. In a copy of the practice script lower it to 1, rerun and inspect timestamps. Restore your copy to 3. For live spaces consult current workspace quotas; do not equate local threads with unlimited server capacity.", "explanation": "To observe actual concurrency and its limits. Expected evidence: Your timestamps demonstrate the difference between serial and parallel execution.", "estimated_time": "2-4 min", "action_type": "verify", "commands": { "windows": "", "macos": "", "linux": "" }, "command_type": "shell", "required_inputs": [], "expected_result": "Your timestamps demonstrate the difference between serial and parallel execution.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Set a quota-aware boundary", "success_condition": "Your timestamps demonstrate the difference between serial and parallel execution.", "how_to_verify": "Record one recognized evidence group: serial, parallel, timestamps. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "Your timestamps demonstrate the difference between serial and parallel execution.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Your timestamps demonstrate the difference between serial and parallel execution.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L06-S02-CHECK", "tokens_any": [ [ "serial", "parallel", "timestamps" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Set a quota-aware boundary, what exact command or inspection failed, and how did its output differ from Your timestamps demonstrate the difference between serial and parallel execution.?", "fix": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Your timestamps demonstrate the difference between serial and parallel execution. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "Why compare serial and parallel timestamps?", "answer": "To observe actual concurrency and its limits. Your timestamps demonstrate the difference between serial and parallel execution." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "Why compare serial and parallel timestamps?", "choices": [ "To assume unlimited server throughput.", "To observe actual concurrency and its limits." ], "correct_answer": 1, "explanation": "To observe actual concurrency and its limits." }, "coach": { "intro": "Next: Set a quota-aware boundary", "demo": "Read the ThreadPoolExecutor max_workers=3 setting. In a copy of the practice script lower it to 1, rerun and inspect timestamps. Restore your copy to 3. For live spaces consult current workspace quotas; do not equate local threads with unlimited server capacity.", "handoff": "Try the action and save what actually happens.", "success": "Your timestamps demonstrate the difference between serial and parallel execution.", "failure": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Your timestamps demonstrate the difference between serial and parallel execution. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Your timestamps demonstrate the difference between serial and parallel execution. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Set a quota-aware boundary", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Concurrency is observed overlap, not a count of worker labels. Workspace quotas can constrain live throughput. Read the ThreadPoolExecutor max_workers=3 setting. In a copy of the practice script lower it to 1, rerun and inspect timestamps. Restore your copy to 3. For live spaces consult current workspace quotas; do not equate local threads with unlimited server capacity. Check: Your timestamps demonstrate the difference between serial and parallel execution.", "on_screen_text": [ "Set a quota-aware boundary", "Your timestamps demonstrate the difference between serial and parallel execution." ], "screen_recording_instructions": [ "Read the ThreadPoolExecutor max_workers=3 setting. In a copy of the practice script lower it to 1, rerun and inspect timestamps. Restore your copy to 3. For live spaces consult current workspace quotas; do not equate local threads with unlimited server capacity.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Your timestamps demonstrate the difference between serial and parallel execution." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Concurrency is observed overlap, not a count of worker labels. Workspace quotas can constrain live throughput.", "Read the ThreadPoolExecutor max_workers=3 setting. In a copy of the practice script lower it to 1, rerun and inspect timestamps. Restore your copy to 3. For live spaces consult current workspace quotas; do not equate local threads with unlimited server capacity.", "Your timestamps demonstrate the difference between serial and parallel execution." ] }, "step_id": "L06-S02", "lesson_id": "L06" }, { "title": "Inject a recoverable failure", "purpose": "Transient failures can justify bounded retry. Invalid data, missing permissions and uncertain remote completion require diagnosis.", "user_action": "Run transient mode. Inspect finance RETRYING in memory and attempt=2 in its result. The call count is four. There is no alternate provider in this small lab; retry uses the same simulated specialist.", "explanation": "Four calls after one bounded retry. Expected evidence: Finance recovers once, all lanes verify and the total call count is 4.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python swarm.py --fault transient", "macos": "python3 swarm.py --fault transient", "linux": "python3 swarm.py --fault transient" }, "command_type": "shell", "required_inputs": [], "expected_result": "Finance recovers once, all lanes verify and the total call count is 4.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Inject a recoverable failure", "success_condition": "Finance recovers once, all lanes verify and the total call count is 4.", "how_to_verify": "Record one recognized evidence group: AWAITING_HUMAN, calls, 4. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "Finance recovers once, all lanes verify and the total call count is 4.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Finance recovers once, all lanes verify and the total call count is 4.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L07-S01-CHECK", "tokens_any": [ [ "AWAITING_HUMAN", "calls", "4" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Inject a recoverable failure, what exact command or inspection failed, and how did its output differ from Finance recovers once, all lanes verify and the total call count is 4.?", "fix": "For this step run python swarm.py --fault transient from the extracted kit folder. Compare against: Finance recovers once, all lanes verify and the total call count is 4. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What is the transient-recovery call count?", "answer": "Four calls after one bounded retry. Finance recovers once, all lanes verify and the total call count is 4." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What is the transient-recovery call count?", "choices": [ "Four calls after one bounded retry.", "Unlimited calls until it works." ], "correct_answer": 0, "explanation": "Four calls after one bounded retry." }, "coach": { "intro": "Next: Inject a recoverable failure", "demo": "Run transient mode. Inspect finance RETRYING in memory and attempt=2 in its result. The call count is four. There is no alternate provider in this small lab; retry uses the same simulated specialist.", "handoff": "Try the action and save what actually happens.", "success": "Finance recovers once, all lanes verify and the total call count is 4.", "failure": "For this step run python swarm.py --fault transient from the extracted kit folder. Compare against: Finance recovers once, all lanes verify and the total call count is 4. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "For this step run python swarm.py --fault transient from the extracted kit folder. Compare against: Finance recovers once, all lanes verify and the total call count is 4. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Inject a recoverable failure", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Transient failures can justify bounded retry. Invalid data, missing permissions and uncertain remote completion require diagnosis. Run transient mode. Inspect finance RETRYING in memory and attempt=2 in its result. The call count is four. There is no alternate provider in this small lab; retry uses the same simulated specialist. Check: Finance recovers once, all lanes verify and the total call count is 4.", "on_screen_text": [ "Inject a recoverable failure", "Finance recovers once, all lanes verify and the total call count is 4." ], "screen_recording_instructions": [ "Run transient mode. Inspect finance RETRYING in memory and attempt=2 in its result. The call count is four. There is no alternate provider in this small lab; retry uses the same simulated specialist.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Finance recovers once, all lanes verify and the total call count is 4." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Transient failures can justify bounded retry. Invalid data, missing permissions and uncertain remote completion require diagnosis.", "Run transient mode. Inspect finance RETRYING in memory and attempt=2 in its result. The call count is four. There is no alternate provider in this small lab; retry uses the same simulated specialist.", "Finance recovers once, all lanes verify and the total call count is 4." ] }, "step_id": "L07-S01", "lesson_id": "L07" }, { "title": "Exhaust recovery safely", "purpose": "Transient failures can justify bounded retry. Invalid data, missing permissions and uncertain remote completion require diagnosis.", "user_action": "Run timeout mode; an exit code of 2 is expected. Compare retained attempts with the call cap. In live mode the code does not automatically resubmit ambiguous timed-out work. Inspect the existing conversation before deciding to retry.", "explanation": "A failed unapproved report and its attempts. Expected evidence: Mission state is FAILED with no approval; the failed evidence remains available.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python swarm.py --fault timeout", "macos": "python3 swarm.py --fault timeout", "linux": "python3 swarm.py --fault timeout" }, "command_type": "shell", "required_inputs": [], "expected_result": "Mission state is FAILED with no approval; the failed evidence remains available.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Exhaust recovery safely", "success_condition": "Mission state is FAILED with no approval; the failed evidence remains available.", "how_to_verify": "Record one recognized evidence group: FAILED, calls. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim. Include run_id: followed by the actual 32 hexadecimal characters from the saved report.", "criteria": [ "Mission state is FAILED with no approval; the failed evidence remains available.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Mission state is FAILED with no approval; the failed evidence remains available.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L07-S02-CHECK", "tokens_any": [ [ "FAILED", "calls" ] ], "deferred_min_reason_chars": 24, "run_id_format": true }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Exhaust recovery safely, what exact command or inspection failed, and how did its output differ from Mission state is FAILED with no approval; the failed evidence remains available.?", "fix": "For this step run python swarm.py --fault timeout from the extracted kit folder. Compare against: Mission state is FAILED with no approval; the failed evidence remains available. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What should an exhausted timeout retain?", "answer": "A failed unapproved report and its attempts. Mission state is FAILED with no approval; the failed evidence remains available." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What should an exhausted timeout retain?", "choices": [ "A success flag inferred from process exit.", "A failed unapproved report and its attempts." ], "correct_answer": 1, "explanation": "A failed unapproved report and its attempts." }, "coach": { "intro": "Next: Exhaust recovery safely", "demo": "Run timeout mode; an exit code of 2 is expected. Compare retained attempts with the call cap. In live mode the code does not automatically resubmit ambiguous timed-out work. Inspect the existing conversation before deciding to retry.", "handoff": "Try the action and save what actually happens.", "success": "Mission state is FAILED with no approval; the failed evidence remains available.", "failure": "For this step run python swarm.py --fault timeout from the extracted kit folder. Compare against: Mission state is FAILED with no approval; the failed evidence remains available. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "For this step run python swarm.py --fault timeout from the extracted kit folder. Compare against: Mission state is FAILED with no approval; the failed evidence remains available. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Exhaust recovery safely", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Transient failures can justify bounded retry. Invalid data, missing permissions and uncertain remote completion require diagnosis. Run timeout mode; an exit code of 2 is expected. Compare retained attempts with the call cap. In live mode the code does not automatically resubmit ambiguous timed-out work. Inspect the existing conversation before deciding to retry. Check: Mission state is FAILED with no approval; the failed evidence remains available.", "on_screen_text": [ "Exhaust recovery safely", "Mission state is FAILED with no approval; the failed evidence remains available." ], "screen_recording_instructions": [ "Run timeout mode; an exit code of 2 is expected. Compare retained attempts with the call cap. In live mode the code does not automatically resubmit ambiguous timed-out work. Inspect the existing conversation before deciding to retry.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Mission state is FAILED with no approval; the failed evidence remains available." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Transient failures can justify bounded retry. Invalid data, missing permissions and uncertain remote completion require diagnosis.", "Run timeout mode; an exit code of 2 is expected. Compare retained attempts with the call cap. In live mode the code does not automatically resubmit ambiguous timed-out work. Inspect the existing conversation before deciding to retry.", "Mission state is FAILED with no approval; the failed evidence remains available." ] }, "step_id": "L07-S02", "lesson_id": "L07" }, { "title": "Reject plausible wrong numbers", "purpose": "An executable oracle compares sample totals independently of the specialist answer. It cannot establish the business meaning of arbitrary production data.", "user_action": "Run bad-data mode. The customer lane returns 999 instead of 3 and the mission fails. Read verify(): column shape, finite numbers, exact sample totals, SQL and source presence.", "explanation": "No; the independent sample oracle must agree. Expected evidence: The customer lane fails verification and the report cannot pass the human gate.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python swarm.py --fault bad-data", "macos": "python3 swarm.py --fault bad-data", "linux": "python3 swarm.py --fault bad-data" }, "command_type": "shell", "required_inputs": [], "expected_result": "The customer lane fails verification and the report cannot pass the human gate.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Reject plausible wrong numbers", "success_condition": "The customer lane fails verification and the report cannot pass the human gate.", "how_to_verify": "Record one recognized evidence group: FAILED, customers. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim. Include run_id: followed by the actual 32 hexadecimal characters from the saved report.", "criteria": [ "The customer lane fails verification and the report cannot pass the human gate.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: The customer lane fails verification and the report cannot pass the human gate.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L08-S01-CHECK", "tokens_any": [ [ "FAILED", "customers" ] ], "deferred_min_reason_chars": 24, "run_id_format": true }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Reject plausible wrong numbers, what exact command or inspection failed, and how did its output differ from The customer lane fails verification and the report cannot pass the human gate.?", "fix": "For this step run python swarm.py --fault bad-data from the extracted kit folder. Compare against: The customer lane fails verification and the report cannot pass the human gate. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "May plausible wrong metric data be released?", "answer": "No; the independent sample oracle must agree. The customer lane fails verification and the report cannot pass the human gate." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "May plausible wrong metric data be released?", "choices": [ "No; the independent sample oracle must agree.", "Yes; well-formatted JSON is enough." ], "correct_answer": 0, "explanation": "No; the independent sample oracle must agree." }, "coach": { "intro": "Next: Reject plausible wrong numbers", "demo": "Run bad-data mode. The customer lane returns 999 instead of 3 and the mission fails. Read verify(): column shape, finite numbers, exact sample totals, SQL and source presence.", "handoff": "Try the action and save what actually happens.", "success": "The customer lane fails verification and the report cannot pass the human gate.", "failure": "For this step run python swarm.py --fault bad-data from the extracted kit folder. Compare against: The customer lane fails verification and the report cannot pass the human gate. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "For this step run python swarm.py --fault bad-data from the extracted kit folder. Compare against: The customer lane fails verification and the report cannot pass the human gate. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Reject plausible wrong numbers", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "An executable oracle compares sample totals independently of the specialist answer. It cannot establish the business meaning of arbitrary production data. Run bad-data mode. The customer lane returns 999 instead of 3 and the mission fails. Read verify(): column shape, finite numbers, exact sample totals, SQL and source presence. Check: The customer lane fails verification and the report cannot pass the human gate.", "on_screen_text": [ "Reject plausible wrong numbers", "The customer lane fails verification and the report cannot pass the human gate." ], "screen_recording_instructions": [ "Run bad-data mode. The customer lane returns 999 instead of 3 and the mission fails. Read verify(): column shape, finite numbers, exact sample totals, SQL and source presence.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "The customer lane fails verification and the report cannot pass the human gate." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "An executable oracle compares sample totals independently of the specialist answer. It cannot establish the business meaning of arbitrary production data.", "Run bad-data mode. The customer lane returns 999 instead of 3 and the mission fails. Read verify(): column shape, finite numbers, exact sample totals, SQL and source presence.", "The customer lane fails verification and the report cannot pass the human gate." ] }, "step_id": "L08-S01", "lesson_id": "L08" }, { "title": "Check source consistency", "purpose": "An executable oracle compares sample totals independently of the specialist answer. It cannot establish the business meaning of arbitrary production data.", "user_action": "Open sample.csv and manually total the three columns. In live SQL use the independent checks from sample.sql. Confirm period, row count and definitions. Do not invent M&A valuation conclusions from three aggregates.", "explanation": "No; a different dataset needs a different oracle. Expected evidence: Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle.", "estimated_time": "2-4 min", "action_type": "verify", "commands": { "windows": "", "macos": "", "linux": "" }, "command_type": "shell", "required_inputs": [], "expected_result": "Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Check source consistency", "success_condition": "Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle.", "how_to_verify": "Record one recognized evidence group: 3000, 3, 4, sample. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L08-S02-CHECK", "tokens_any": [ [ "3000", "3", "4", "sample" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Check source consistency, what exact command or inspection failed, and how did its output differ from Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle.?", "fix": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "Can sample totals verify another business dataset?", "answer": "No; a different dataset needs a different oracle. Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "Can sample totals verify another business dataset?", "choices": [ "Yes; use the same three totals.", "No; a different dataset needs a different oracle." ], "correct_answer": 1, "explanation": "No; a different dataset needs a different oracle." }, "coach": { "intro": "Next: Check source consistency", "demo": "Open sample.csv and manually total the three columns. In live SQL use the independent checks from sample.sql. Confirm period, row count and definitions. Do not invent M&A valuation conclusions from three aggregates.", "handoff": "Try the action and save what actually happens.", "success": "Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle.", "failure": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Check source consistency", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "An executable oracle compares sample totals independently of the specialist answer. It cannot establish the business meaning of arbitrary production data. Open sample.csv and manually total the three columns. In live SQL use the independent checks from sample.sql. Confirm period, row count and definitions. Do not invent M&A valuation conclusions from three aggregates. Check: Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle.", "on_screen_text": [ "Check source consistency", "Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle." ], "screen_recording_instructions": [ "Open sample.csv and manually total the three columns. In live SQL use the independent checks from sample.sql. Confirm period, row count and definitions. Do not invent M&A valuation conclusions from three aggregates.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "An executable oracle compares sample totals independently of the specialist answer. It cannot establish the business meaning of arbitrary production data.", "Open sample.csv and manually total the three columns. In live SQL use the independent checks from sample.sql. Confirm period, row count and definitions. Do not invent M&A valuation conclusions from three aggregates.", "Both sample rows produce 3000 revenue, 3 churned and 4 late orders; a different dataset needs a different oracle." ] }, "step_id": "L08-S02", "lesson_id": "L08" }, { "title": "Open the dashboard", "purpose": "The dashboard shows actual retained reports and local approvals. It runs practice only; no token count or estimated DBU value is fabricated.", "user_action": "After installing requirements, run this command in the extracted folder. Open http://127.0.0.1:8082 and run a sample mission. The button performs a short practice run; this dashboard shows the result after completion, not streaming worker telemetry. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution. If your assigned example port 8082 is occupied, use --server.port 8092 instead; if your fleet identity differs, use its allocated port from safety.md. Keep --server.address 127.0.0.1. Never borrow another agent’s assigned port.", "explanation": "Retained practice reports with a hash and approval state. Expected evidence: PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python -m streamlit run app.py --server.address 127.0.0.1 --server.port 8082", "macos": "python3 -m streamlit run app.py --server.address 127.0.0.1 --server.port 8082", "linux": "python3 -m streamlit run app.py --server.address 127.0.0.1 --server.port 8082" }, "command_type": "shell", "required_inputs": [], "expected_result": "PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Open the dashboard", "success_condition": "PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible.", "how_to_verify": "Record one recognized evidence group: PRACTICE, AWAITING_HUMAN, hash OR DEFERRED, reason. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim. Include run_id: followed by the actual 32 hexadecimal characters from the saved report. Include REPORT_SHA256: followed by the actual 64 hexadecimal characters. Format checks cannot authenticate a cloud run.", "criteria": [ "PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L09-S01-CHECK", "tokens_any": [ [ "PRACTICE", "AWAITING_HUMAN", "hash" ], [ "DEFERRED", "reason" ] ], "deferred_min_reason_chars": 24, "run_id_format": true, "report_hash_format": true }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Open the dashboard, what exact command or inspection failed, and how did its output differ from PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible.?", "fix": "For this step run python -m streamlit run app.py --server.address 127.0.0.1 --server.port 8082 from the extracted kit folder. Compare against: PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible. ModuleNotFoundError means install the listed requirements in this Python environment. An occupied port needs your assigned fallback loopback port. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What does the required dashboard show?", "answer": "Retained practice reports with a hash and approval state. PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What does the required dashboard show?", "choices": [ "Retained practice reports with a hash and approval state.", "Certified live DBU spending and token meters." ], "correct_answer": 0, "explanation": "Retained practice reports with a hash and approval state." }, "coach": { "intro": "Next: Open the dashboard", "demo": "After installing requirements, run this command in the extracted folder. Open http://127.0.0.1:8082 and run a sample mission. The button performs a short practice run; this dashboard shows the result after completion, not streaming worker telemetry. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution. If your assigned example port 8082 is occupied, use --server.port 8092 instead; if your fleet identity differs, use its allocated port from safety.md. Keep --server.address 127.0.0.1. Never borrow another agent’s assigned port.", "handoff": "Try the action and save what actually happens.", "success": "PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible.", "failure": "For this step run python -m streamlit run app.py --server.address 127.0.0.1 --server.port 8082 from the extracted kit folder. Compare against: PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible. ModuleNotFoundError means install the listed requirements in this Python environment. An occupied port needs your assigned fallback loopback port. ", "help": "For this step run python -m streamlit run app.py --server.address 127.0.0.1 --server.port 8082 from the extracted kit folder. Compare against: PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible. ModuleNotFoundError means install the listed requirements in this Python environment. An occupied port needs your assigned fallback loopback port. " }, "video": { "purpose": "Open the dashboard", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "The dashboard shows actual retained reports and local approvals. It runs practice only; no token count or estimated DBU value is fabricated. After installing requirements, run this command in the extracted folder. Open http://127.0.0.1:8082 and run a sample mission. The button performs a short practice run; this dashboard shows the result after completion, not streaming worker telemetry. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution. If your assigned example port 8082 is occupied, use --server.port 8092 instead; if your fleet identity differs, use its allocated port from safety.md. Keep --server.address 127.0.0.1. Never borrow another agent’s assigned port. Check: PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible.", "on_screen_text": [ "Open the dashboard", "PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible." ], "screen_recording_instructions": [ "After installing requirements, run this command in the extracted folder. Open http://127.0.0.1:8082 and run a sample mission. The button performs a short practice run; this dashboard shows the result after completion, not streaming worker telemetry. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution. If your assigned example port 8082 is occupied, use --server.port 8092 instead; if your fleet identity differs, use its allocated port from safety.md. Keep --server.address 127.0.0.1. Never borrow another agent’s assigned port.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "The dashboard shows actual retained reports and local approvals. It runs practice only; no token count or estimated DBU value is fabricated.", "After installing requirements, run this command in the extracted folder. Open http://127.0.0.1:8082 and run a sample mission. The button performs a short practice run; this dashboard shows the result after completion, not streaming worker telemetry. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution. If your assigned example port 8082 is occupied, use --server.port 8092 instead; if your fleet identity differs, use its allocated port from safety.md. Keep --server.address 127.0.0.1. Never borrow another agent’s assigned port.", "PRACTICE label, AWAITING_HUMAN, report bytes, hash and retained history are visible." ] }, "step_id": "L09-S01", "lesson_id": "L09" }, { "title": "Review a failed mission", "purpose": "The dashboard shows actual retained reports and local approvals. It runs practice only; no token count or estimated DBU value is fabricated.", "user_action": "Choose bad-data in the dashboard and run. Confirm FAILED and no approval button. Read cost: only practice has 0; live cost is unknown until billing evidence is collected. Stop Streamlit with Ctrl+C when finished.", "explanation": "None; failed reports cannot be approved. Expected evidence: An invalid report has no release action and no invented cloud cost meter.", "estimated_time": "2-4 min", "action_type": "verify", "commands": { "windows": "", "macos": "", "linux": "" }, "command_type": "shell", "required_inputs": [], "expected_result": "An invalid report has no release action and no invented cloud cost meter.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Review a failed mission", "success_condition": "An invalid report has no release action and no invented cloud cost meter.", "how_to_verify": "Record one recognized evidence group: FAILED, approval. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim. Include run_id: followed by the actual 32 hexadecimal characters from the saved report.", "criteria": [ "An invalid report has no release action and no invented cloud cost meter.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: An invalid report has no release action and no invented cloud cost meter.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L09-S02-CHECK", "tokens_any": [ [ "FAILED", "approval" ] ], "deferred_min_reason_chars": 24, "run_id_format": true }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Review a failed mission, what exact command or inspection failed, and how did its output differ from An invalid report has no release action and no invented cloud cost meter.?", "fix": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: An invalid report has no release action and no invented cloud cost meter. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What release action should an invalid report have?", "answer": "None; failed reports cannot be approved. An invalid report has no release action and no invented cloud cost meter." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What release action should an invalid report have?", "choices": [ "Approve if it looks plausible.", "None; failed reports cannot be approved." ], "correct_answer": 1, "explanation": "None; failed reports cannot be approved." }, "coach": { "intro": "Next: Review a failed mission", "demo": "Choose bad-data in the dashboard and run. Confirm FAILED and no approval button. Read cost: only practice has 0; live cost is unknown until billing evidence is collected. Stop Streamlit with Ctrl+C when finished.", "handoff": "Try the action and save what actually happens.", "success": "An invalid report has no release action and no invented cloud cost meter.", "failure": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: An invalid report has no release action and no invented cloud cost meter. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: An invalid report has no release action and no invented cloud cost meter. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Review a failed mission", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "The dashboard shows actual retained reports and local approvals. It runs practice only; no token count or estimated DBU value is fabricated. Choose bad-data in the dashboard and run. Confirm FAILED and no approval button. Read cost: only practice has 0; live cost is unknown until billing evidence is collected. Stop Streamlit with Ctrl+C when finished. Check: An invalid report has no release action and no invented cloud cost meter.", "on_screen_text": [ "Review a failed mission", "An invalid report has no release action and no invented cloud cost meter." ], "screen_recording_instructions": [ "Choose bad-data in the dashboard and run. Confirm FAILED and no approval button. Read cost: only practice has 0; live cost is unknown until billing evidence is collected. Stop Streamlit with Ctrl+C when finished.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "An invalid report has no release action and no invented cloud cost meter." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "The dashboard shows actual retained reports and local approvals. It runs practice only; no token count or estimated DBU value is fabricated.", "Choose bad-data in the dashboard and run. Confirm FAILED and no approval button. Read cost: only practice has 0; live cost is unknown until billing evidence is collected. Stop Streamlit with Ctrl+C when finished.", "An invalid report has no release action and no invented cloud cost meter." ] }, "step_id": "L09-S02", "lesson_id": "L09" }, { "title": "Trip a call boundary", "purpose": "Local call and runtime checks stop new work. Remote queries may outlive client timeouts; hard cloud limits require administrator-enforced controls.", "user_action": "Run with one call. Inspect FAILED lanes. Record max calls, SQL warehouse auto-stop, billing monitoring and a named owner for live spending. Budgets/alerts are not automatically hard spend stops. No autonomous external writes or publication are enabled.", "explanation": "Before another dispatch exceeds the limit. Expected evidence: At most one specialist call occurs and the mission fails closed.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python swarm.py --max-calls 1", "macos": "python3 swarm.py --max-calls 1", "linux": "python3 swarm.py --max-calls 1" }, "command_type": "shell", "required_inputs": [], "expected_result": "At most one specialist call occurs and the mission fails closed.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Trip a call boundary", "success_condition": "At most one specialist call occurs and the mission fails closed.", "how_to_verify": "Record one recognized evidence group: FAILED, calls, 1. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim. Include run_id: followed by the actual 32 hexadecimal characters from the saved report.", "criteria": [ "At most one specialist call occurs and the mission fails closed.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: At most one specialist call occurs and the mission fails closed.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L10-S01-CHECK", "tokens_any": [ [ "FAILED", "calls", "1" ] ], "deferred_min_reason_chars": 24, "run_id_format": true }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Trip a call boundary, what exact command or inspection failed, and how did its output differ from At most one specialist call occurs and the mission fails closed.?", "fix": "For this step run python swarm.py --max-calls 1 from the extracted kit folder. Compare against: At most one specialist call occurs and the mission fails closed. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "When should the call ceiling stop work?", "answer": "Before another dispatch exceeds the limit. At most one specialist call occurs and the mission fails closed." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "When should the call ceiling stop work?", "choices": [ "Before another dispatch exceeds the limit.", "After dispatch, for accounting only." ], "correct_answer": 0, "explanation": "Before another dispatch exceeds the limit." }, "coach": { "intro": "Next: Trip a call boundary", "demo": "Run with one call. Inspect FAILED lanes. Record max calls, SQL warehouse auto-stop, billing monitoring and a named owner for live spending. Budgets/alerts are not automatically hard spend stops. No autonomous external writes or publication are enabled.", "handoff": "Try the action and save what actually happens.", "success": "At most one specialist call occurs and the mission fails closed.", "failure": "For this step run python swarm.py --max-calls 1 from the extracted kit folder. Compare against: At most one specialist call occurs and the mission fails closed. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "For this step run python swarm.py --max-calls 1 from the extracted kit folder. Compare against: At most one specialist call occurs and the mission fails closed. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Trip a call boundary", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Local call and runtime checks stop new work. Remote queries may outlive client timeouts; hard cloud limits require administrator-enforced controls. Run with one call. Inspect FAILED lanes. Record max calls, SQL warehouse auto-stop, billing monitoring and a named owner for live spending. Budgets/alerts are not automatically hard spend stops. No autonomous external writes or publication are enabled. Check: At most one specialist call occurs and the mission fails closed.", "on_screen_text": [ "Trip a call boundary", "At most one specialist call occurs and the mission fails closed." ], "screen_recording_instructions": [ "Run with one call. Inspect FAILED lanes. Record max calls, SQL warehouse auto-stop, billing monitoring and a named owner for live spending. Budgets/alerts are not automatically hard spend stops. No autonomous external writes or publication are enabled.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "At most one specialist call occurs and the mission fails closed." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Local call and runtime checks stop new work. Remote queries may outlive client timeouts; hard cloud limits require administrator-enforced controls.", "Run with one call. Inspect FAILED lanes. Record max calls, SQL warehouse auto-stop, billing monitoring and a named owner for live spending. Budgets/alerts are not automatically hard spend stops. No autonomous external writes or publication are enabled.", "At most one specialist call occurs and the mission fails closed." ] }, "step_id": "L10-S01", "lesson_id": "L10" }, { "title": "Approve exact reviewed bytes", "purpose": "Local call and runtime checks stop new work. Remote queries may outlive client timeouts; hard cloud limits require administrator-enforced controls.", "user_action": "Run a successful practice mission. Copy its run ID and REPORT_SHA256. Replace RUN_ID and SHA256 in the command, review that JSON and approve it. Then change a metric in a copy and repeat with the old hash: rejection is required. This gate is local evidence, not an authenticated enterprise approver service. Replace runs/RUN_ID.json with the actual saved report path and SHA256 with the printed REPORT_SHA256 after reading the report. A literal placeholder is rejected with exit 1.", "explanation": "The exact verified report hash. Expected evidence: A receipt binds the unchanged report hash; altered or failed reports are rejected.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python swarm.py --approve runs/RUN_ID.json --hash SHA256", "macos": "python3 swarm.py --approve runs/RUN_ID.json --hash SHA256", "linux": "python3 swarm.py --approve runs/RUN_ID.json --hash SHA256" }, "command_type": "shell", "required_inputs": [], "expected_result": "A receipt binds the unchanged report hash; altered or failed reports are rejected.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Approve exact reviewed bytes", "success_condition": "A receipt binds the unchanged report hash; altered or failed reports are rejected.", "how_to_verify": "Record one recognized evidence group: human-approved-local-artifact, report_sha256. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim. Include run_id: followed by the actual 32 hexadecimal characters from the saved report. Include REPORT_SHA256: followed by the actual 64 hexadecimal characters. Format checks cannot authenticate a cloud run.", "criteria": [ "A receipt binds the unchanged report hash; altered or failed reports are rejected.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: A receipt binds the unchanged report hash; altered or failed reports are rejected.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L10-S02-CHECK", "tokens_any": [ [ "human-approved-local-artifact", "report_sha256" ] ], "deferred_min_reason_chars": 24, "run_id_format": true, "report_hash_format": true }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Approve exact reviewed bytes, what exact command or inspection failed, and how did its output differ from A receipt binds the unchanged report hash; altered or failed reports are rejected.?", "fix": "For this step run python swarm.py --approve runs/RUN_ID.json --hash SHA256 from the extracted kit folder. Compare against: A receipt binds the unchanged report hash; altered or failed reports are rejected. Replace RUN_ID/report path and SHA256 with this retained verified report; a literal placeholder or changed report must be rejected. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What binds approval to the reviewed content?", "answer": "The exact verified report hash. A receipt binds the unchanged report hash; altered or failed reports are rejected." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What binds approval to the reviewed content?", "choices": [ "A filename remembered from an earlier draft.", "The exact verified report hash." ], "correct_answer": 1, "explanation": "The exact verified report hash." }, "coach": { "intro": "Next: Approve exact reviewed bytes", "demo": "Run a successful practice mission. Copy its run ID and REPORT_SHA256. Replace RUN_ID and SHA256 in the command, review that JSON and approve it. Then change a metric in a copy and repeat with the old hash: rejection is required. This gate is local evidence, not an authenticated enterprise approver service. Replace runs/RUN_ID.json with the actual saved report path and SHA256 with the printed REPORT_SHA256 after reading the report. A literal placeholder is rejected with exit 1.", "handoff": "Try the action and save what actually happens.", "success": "A receipt binds the unchanged report hash; altered or failed reports are rejected.", "failure": "For this step run python swarm.py --approve runs/RUN_ID.json --hash SHA256 from the extracted kit folder. Compare against: A receipt binds the unchanged report hash; altered or failed reports are rejected. Replace RUN_ID/report path and SHA256 with this retained verified report; a literal placeholder or changed report must be rejected. ", "help": "For this step run python swarm.py --approve runs/RUN_ID.json --hash SHA256 from the extracted kit folder. Compare against: A receipt binds the unchanged report hash; altered or failed reports are rejected. Replace RUN_ID/report path and SHA256 with this retained verified report; a literal placeholder or changed report must be rejected. " }, "video": { "purpose": "Approve exact reviewed bytes", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Local call and runtime checks stop new work. Remote queries may outlive client timeouts; hard cloud limits require administrator-enforced controls. Run a successful practice mission. Copy its run ID and REPORT_SHA256. Replace RUN_ID and SHA256 in the command, review that JSON and approve it. Then change a metric in a copy and repeat with the old hash: rejection is required. This gate is local evidence, not an authenticated enterprise approver service. Replace runs/RUN_ID.json with the actual saved report path and SHA256 with the printed REPORT_SHA256 after reading the report. A literal placeholder is rejected with exit 1. Check: A receipt binds the unchanged report hash; altered or failed reports are rejected.", "on_screen_text": [ "Approve exact reviewed bytes", "A receipt binds the unchanged report hash; altered or failed reports are rejected." ], "screen_recording_instructions": [ "Run a successful practice mission. Copy its run ID and REPORT_SHA256. Replace RUN_ID and SHA256 in the command, review that JSON and approve it. Then change a metric in a copy and repeat with the old hash: rejection is required. This gate is local evidence, not an authenticated enterprise approver service. Replace runs/RUN_ID.json with the actual saved report path and SHA256 with the printed REPORT_SHA256 after reading the report. A literal placeholder is rejected with exit 1.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "A receipt binds the unchanged report hash; altered or failed reports are rejected." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Local call and runtime checks stop new work. Remote queries may outlive client timeouts; hard cloud limits require administrator-enforced controls.", "Run a successful practice mission. Copy its run ID and REPORT_SHA256. Replace RUN_ID and SHA256 in the command, review that JSON and approve it. Then change a metric in a copy and repeat with the old hash: rejection is required. This gate is local evidence, not an authenticated enterprise approver service. Replace runs/RUN_ID.json with the actual saved report path and SHA256 with the printed REPORT_SHA256 after reading the report. A literal placeholder is rejected with exit 1.", "A receipt binds the unchanged report hash; altered or failed reports are rejected." ] }, "step_id": "L10-S02", "lesson_id": "L10" }, { "title": "Repeat from a fresh run", "purpose": "Revenue leakage and churn questions can coordinate specialists; causal explanations need more evidence than aggregates.", "user_action": "Run the normal command again. Compare run IDs, inspect all three source-bearing results and record limitations. Retain prior runs and errors. No repeated-run proof should overwrite the first experiment.", "explanation": "A unique retained run ID and new evidence. Expected evidence: A fresh unique run with three checked metrics and an unapproved final artifact is retained.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python swarm.py", "macos": "python3 swarm.py", "linux": "python3 swarm.py" }, "command_type": "shell", "required_inputs": [], "expected_result": "A fresh unique run with three checked metrics and an unapproved final artifact is retained.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Repeat from a fresh run", "success_condition": "A fresh unique run with three checked metrics and an unapproved final artifact is retained.", "how_to_verify": "Record one recognized evidence group: AWAITING_HUMAN, run_id, REPORT_SHA256. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim. Include run_id: followed by the actual 32 hexadecimal characters from the saved report. Include REPORT_SHA256: followed by the actual 64 hexadecimal characters. Format checks cannot authenticate a cloud run.", "criteria": [ "A fresh unique run with three checked metrics and an unapproved final artifact is retained.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: A fresh unique run with three checked metrics and an unapproved final artifact is retained.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L11-S01-CHECK", "tokens_any": [ [ "AWAITING_HUMAN", "run_id", "REPORT_SHA256" ] ], "deferred_min_reason_chars": 24, "run_id_format": true, "report_hash_format": true }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Repeat from a fresh run, what exact command or inspection failed, and how did its output differ from A fresh unique run with three checked metrics and an unapproved final artifact is retained.?", "fix": "For this step run python swarm.py from the extracted kit folder. Compare against: A fresh unique run with three checked metrics and an unapproved final artifact is retained. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What distinguishes a fresh run?", "answer": "A unique retained run ID and new evidence. A fresh unique run with three checked metrics and an unapproved final artifact is retained." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What distinguishes a fresh run?", "choices": [ "A unique retained run ID and new evidence.", "Overwriting yesterday’s results." ], "correct_answer": 0, "explanation": "A unique retained run ID and new evidence." }, "coach": { "intro": "Next: Repeat from a fresh run", "demo": "Run the normal command again. Compare run IDs, inspect all three source-bearing results and record limitations. Retain prior runs and errors. No repeated-run proof should overwrite the first experiment.", "handoff": "Try the action and save what actually happens.", "success": "A fresh unique run with three checked metrics and an unapproved final artifact is retained.", "failure": "For this step run python swarm.py from the extracted kit folder. Compare against: A fresh unique run with three checked metrics and an unapproved final artifact is retained. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "For this step run python swarm.py from the extracted kit folder. Compare against: A fresh unique run with three checked metrics and an unapproved final artifact is retained. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Repeat from a fresh run", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Revenue leakage and churn questions can coordinate specialists; causal explanations need more evidence than aggregates. Run the normal command again. Compare run IDs, inspect all three source-bearing results and record limitations. Retain prior runs and errors. No repeated-run proof should overwrite the first experiment. Check: A fresh unique run with three checked metrics and an unapproved final artifact is retained.", "on_screen_text": [ "Repeat from a fresh run", "A fresh unique run with three checked metrics and an unapproved final artifact is retained." ], "screen_recording_instructions": [ "Run the normal command again. Compare run IDs, inspect all three source-bearing results and record limitations. Retain prior runs and errors. No repeated-run proof should overwrite the first experiment.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "A fresh unique run with three checked metrics and an unapproved final artifact is retained." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Revenue leakage and churn questions can coordinate specialists; causal explanations need more evidence than aggregates.", "Run the normal command again. Compare run IDs, inspect all three source-bearing results and record limitations. Retain prior runs and errors. No repeated-run proof should overwrite the first experiment.", "A fresh unique run with three checked metrics and an unapproved final artifact is retained." ] }, "step_id": "L11-S01", "lesson_id": "L11" }, { "title": "Write an evidence-based conclusion", "purpose": "Revenue leakage and churn questions can coordinate specialists; causal explanations need more evidence than aggregates.", "user_action": "In evidence.md write the totals, sources, verification method and uncertainty. State that this tiny sample does not establish causes, forecasts, company value or a general autonomous planner. Review before approving the local report.", "explanation": "Only claims supported by the observed results. Expected evidence: Every factual claim maps to a result; unsupported strategic conclusions remain unanswered.", "estimated_time": "2-4 min", "action_type": "verify", "commands": { "windows": "", "macos": "", "linux": "" }, "command_type": "shell", "required_inputs": [], "expected_result": "Every factual claim maps to a result; unsupported strategic conclusions remain unanswered.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Write an evidence-based conclusion", "success_condition": "Every factual claim maps to a result; unsupported strategic conclusions remain unanswered.", "how_to_verify": "Record one recognized evidence group: claim, result, unsupported. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "Every factual claim maps to a result; unsupported strategic conclusions remain unanswered.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Every factual claim maps to a result; unsupported strategic conclusions remain unanswered.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L11-S02-CHECK", "tokens_any": [ [ "claim", "result", "unsupported" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Write an evidence-based conclusion, what exact command or inspection failed, and how did its output differ from Every factual claim maps to a result; unsupported strategic conclusions remain unanswered.?", "fix": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Every factual claim maps to a result; unsupported strategic conclusions remain unanswered. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "Which business conclusions may the report assert?", "answer": "Only claims supported by the observed results. Every factual claim maps to a result; unsupported strategic conclusions remain unanswered." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "Which business conclusions may the report assert?", "choices": [ "Any causal strategy that sounds reasonable.", "Only claims supported by the observed results." ], "correct_answer": 1, "explanation": "Only claims supported by the observed results." }, "coach": { "intro": "Next: Write an evidence-based conclusion", "demo": "In evidence.md write the totals, sources, verification method and uncertainty. State that this tiny sample does not establish causes, forecasts, company value or a general autonomous planner. Review before approving the local report.", "handoff": "Try the action and save what actually happens.", "success": "Every factual claim maps to a result; unsupported strategic conclusions remain unanswered.", "failure": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Every factual claim maps to a result; unsupported strategic conclusions remain unanswered. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Every factual claim maps to a result; unsupported strategic conclusions remain unanswered. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Write an evidence-based conclusion", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Revenue leakage and churn questions can coordinate specialists; causal explanations need more evidence than aggregates. In evidence.md write the totals, sources, verification method and uncertainty. State that this tiny sample does not establish causes, forecasts, company value or a general autonomous planner. Review before approving the local report. Check: Every factual claim maps to a result; unsupported strategic conclusions remain unanswered.", "on_screen_text": [ "Write an evidence-based conclusion", "Every factual claim maps to a result; unsupported strategic conclusions remain unanswered." ], "screen_recording_instructions": [ "In evidence.md write the totals, sources, verification method and uncertainty. State that this tiny sample does not establish causes, forecasts, company value or a general autonomous planner. Review before approving the local report.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Every factual claim maps to a result; unsupported strategic conclusions remain unanswered." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Revenue leakage and churn questions can coordinate specialists; causal explanations need more evidence than aggregates.", "In evidence.md write the totals, sources, verification method and uncertainty. State that this tiny sample does not establish causes, forecasts, company value or a general autonomous planner. Review before approving the local report.", "Every factual claim maps to a result; unsupported strategic conclusions remain unanswered." ] }, "step_id": "L11-S02", "lesson_id": "L11" }, { "title": "Prepare a Databricks App", "purpose": "Cloud deployment requires your explicit account authorization and permissions. A service principal needs separate grants; an app definition alone grants none.", "user_action": "Read deployment.md. Inspect app.yaml and requirements.txt, test locally first and copy only reviewed app resources to an approved workspace folder. Create an app in the workspace Apps UI and deploy that folder after checking cost and sharing. This app remains practice-only. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "explanation": "No; permissions, deployment and durable storage need live evidence. Expected evidence: The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred.", "estimated_time": "2-4 min", "action_type": "verify", "commands": { "windows": "", "macos": "", "linux": "" }, "command_type": "shell", "required_inputs": [], "expected_result": "The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Prepare a Databricks App", "success_condition": "The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred.", "how_to_verify": "Record one recognized evidence group: Apps, permissions, storage OR DEFERRED, reason. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L12-S01-CHECK", "tokens_any": [ [ "Apps", "permissions", "storage" ], [ "DEFERRED", "reason" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Prepare a Databricks App, what exact command or inspection failed, and how did its output differ from The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred.?", "fix": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "Does a local app template certify cloud deployment?", "answer": "No; permissions, deployment and durable storage need live evidence. The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "Does a local app template certify cloud deployment?", "choices": [ "No; permissions, deployment and durable storage need live evidence.", "Yes; the yaml file is a deployed app." ], "correct_answer": 0, "explanation": "No; permissions, deployment and durable storage need live evidence." }, "coach": { "intro": "Next: Prepare a Databricks App", "demo": "Read deployment.md. Inspect app.yaml and requirements.txt, test locally first and copy only reviewed app resources to an approved workspace folder. Create an app in the workspace Apps UI and deploy that folder after checking cost and sharing. This app remains practice-only. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "handoff": "Try the action and save what actually happens.", "success": "The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred.", "failure": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Prepare a Databricks App", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Cloud deployment requires your explicit account authorization and permissions. A service principal needs separate grants; an app definition alone grants none. Read deployment.md. Inspect app.yaml and requirements.txt, test locally first and copy only reviewed app resources to an approved workspace folder. Create an app in the workspace Apps UI and deploy that folder after checking cost and sharing. This app remains practice-only. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution. Check: The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred.", "on_screen_text": [ "Prepare a Databricks App", "The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred." ], "screen_recording_instructions": [ "Read deployment.md. Inspect app.yaml and requirements.txt, test locally first and copy only reviewed app resources to an approved workspace folder. Create an app in the workspace Apps UI and deploy that folder after checking cost and sharing. This app remains practice-only. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Cloud deployment requires your explicit account authorization and permissions. A service principal needs separate grants; an app definition alone grants none.", "Read deployment.md. Inspect app.yaml and requirements.txt, test locally first and copy only reviewed app resources to an approved workspace folder. Create an app in the workspace Apps UI and deploy that folder after checking cost and sharing. This app remains practice-only. If this optional live or dependency check is unavailable, enter DEFERRED with a concrete reason; continue the free practice track without claiming live execution.", "The app configuration and required permissions are reviewed; deployment is verified in a workspace or marked deferred." ] }, "step_id": "L12-S01", "lesson_id": "L12" }, { "title": "Choose the current agent route", "purpose": "Cloud deployment requires your explicit account authorization and permissions. A service principal needs separate grants; an app definition alone grants none.", "user_action": "Read current-route.md and the linked official docs. The October 12 docs label old Supervisor Agent and agent-server templates legacy. For new live multi-agent AI use current Genie Agents/Agent Bricks CLI guidance; do not copy a legacy deployment as a new recommended architecture. Rerun extracted-kit tests before release.", "explanation": "Observed practice acceptance with separate live limits. Expected evidence: Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python test_swarm.py", "macos": "python3 test_swarm.py", "linux": "python3 test_swarm.py" }, "command_type": "shell", "required_inputs": [], "expected_result": "Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Choose the current agent route", "success_condition": "Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate.", "how_to_verify": "Record one recognized evidence group: PASS: parallel overlap. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L12-S02-CHECK", "tokens_any": [ [ "PASS: parallel overlap" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Choose the current agent route, what exact command or inspection failed, and how did its output differ from Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate.?", "fix": "For this step run python test_swarm.py from the extracted kit folder. Compare against: Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What is accepted on the free completion track?", "answer": "Observed practice acceptance with separate live limits. Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What is accepted on the free completion track?", "choices": [ "A claim of autonomous cloud graduation.", "Observed practice acceptance with separate live limits." ], "correct_answer": 1, "explanation": "Observed practice acceptance with separate live limits." }, "coach": { "intro": "Next: Choose the current agent route", "demo": "Read current-route.md and the linked official docs. The October 12 docs label old Supervisor Agent and agent-server templates legacy. For new live multi-agent AI use current Genie Agents/Agent Bricks CLI guidance; do not copy a legacy deployment as a new recommended architecture. Rerun extracted-kit tests before release.", "handoff": "Try the action and save what actually happens.", "success": "Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate.", "failure": "For this step run python test_swarm.py from the extracted kit folder. Compare against: Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "For this step run python test_swarm.py from the extracted kit folder. Compare against: Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Choose the current agent route", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Cloud deployment requires your explicit account authorization and permissions. A service principal needs separate grants; an app definition alone grants none. Read current-route.md and the linked official docs. The October 12 docs label old Supervisor Agent and agent-server templates legacy. For new live multi-agent AI use current Genie Agents/Agent Bricks CLI guidance; do not copy a legacy deployment as a new recommended architecture. Rerun extracted-kit tests before release. Check: Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate.", "on_screen_text": [ "Choose the current agent route", "Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate." ], "screen_recording_instructions": [ "Read current-route.md and the linked official docs. The October 12 docs label old Supervisor Agent and agent-server templates legacy. For new live multi-agent AI use current Genie Agents/Agent Bricks CLI guidance; do not copy a legacy deployment as a new recommended architecture. Rerun extracted-kit tests before release.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Cloud deployment requires your explicit account authorization and permissions. A service principal needs separate grants; an app definition alone grants none.", "Read current-route.md and the linked official docs. The October 12 docs label old Supervisor Agent and agent-server templates legacy. For new live multi-agent AI use current Genie Agents/Agent Bricks CLI guidance; do not copy a legacy deployment as a new recommended architecture. Rerun extracted-kit tests before release.", "Practice acceptance passes and live deployment/modern orchestration evidence is clearly separate." ] }, "step_id": "L12-S02", "lesson_id": "L12" }, { "title": "Attempt executive due diligence", "purpose": "Graduation requires a new mission and honest scope. Durable memory, owner identity and reproducible failure evidence matter more than a self-awarded grade.", "user_action": "Read capstone.md. For practice repeat the constrained three-metric due-diligence rehearsal and record its limited scope. For live graduation supply a new authorized dataset, independently checked questions/oracles, actual three-space results, failure handling, limits and approval. This shipped fixed planner does not solve arbitrary valuation challenges.", "explanation": "Sample rehearsal is distinct from real M&A autonomy. Expected evidence: Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation.", "estimated_time": "2-4 min", "action_type": "verify", "commands": { "windows": "", "macos": "", "linux": "" }, "command_type": "shell", "required_inputs": [], "expected_result": "Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Attempt executive due diligence", "success_condition": "Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation.", "how_to_verify": "Record one recognized evidence group: PRACTICE, M&A, unearned. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim.", "criteria": [ "Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L13-S01-CHECK", "tokens_any": [ [ "PRACTICE", "M&A", "unearned" ] ], "deferred_min_reason_chars": 24 }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Attempt executive due diligence, what exact command or inspection failed, and how did its output differ from Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation.?", "fix": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What must an executive due-diligence capstone disclose?", "answer": "Sample rehearsal is distinct from real M&A autonomy. Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What must an executive due-diligence capstone disclose?", "choices": [ "Sample rehearsal is distinct from real M&A autonomy.", "Sample math proves investment conclusions." ], "correct_answer": 0, "explanation": "Sample rehearsal is distinct from real M&A autonomy." }, "coach": { "intro": "Next: Attempt executive due diligence", "demo": "Read capstone.md. For practice repeat the constrained three-metric due-diligence rehearsal and record its limited scope. For live graduation supply a new authorized dataset, independently checked questions/oracles, actual three-space results, failure handling, limits and approval. This shipped fixed planner does not solve arbitrary valuation challenges.", "handoff": "Try the action and save what actually happens.", "success": "Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation.", "failure": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "This is an inspection step, not a terminal command. Reopen the diagram or artifact named in the action. Compare against: Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Attempt executive due diligence", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Graduation requires a new mission and honest scope. Durable memory, owner identity and reproducible failure evidence matter more than a self-awarded grade. Read capstone.md. For practice repeat the constrained three-metric due-diligence rehearsal and record its limited scope. For live graduation supply a new authorized dataset, independently checked questions/oracles, actual three-space results, failure handling, limits and approval. This shipped fixed planner does not solve arbitrary valuation challenges. Check: Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation.", "on_screen_text": [ "Attempt executive due diligence", "Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation." ], "screen_recording_instructions": [ "Read capstone.md. For practice repeat the constrained three-metric due-diligence rehearsal and record its limited scope. For live graduation supply a new authorized dataset, independently checked questions/oracles, actual three-space results, failure handling, limits and approval. This shipped fixed planner does not solve arbitrary valuation challenges.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Graduation requires a new mission and honest scope. Durable memory, owner identity and reproducible failure evidence matter more than a self-awarded grade.", "Read capstone.md. For practice repeat the constrained three-metric due-diligence rehearsal and record its limited scope. For live graduation supply a new authorized dataset, independently checked questions/oracles, actual three-space results, failure handling, limits and approval. This shipped fixed planner does not solve arbitrary valuation challenges.", "Your capstone record distinguishes the completed sample rehearsal from unearned general autonomous M&A graduation." ] }, "step_id": "L13-S01", "lesson_id": "L13" }, { "title": "Build Bot Better and run later", "purpose": "Graduation requires a new mission and honest scope. Durable memory, owner identity and reproducible failure evidence matter more than a self-awarded grade.", "user_action": "Read safety.md. Save reports, config without tokens and a copy of memory.sqlite while the process is stopped; restore to a new folder and inspect. Onboard each real worker under its own login, folder and grants; never borrow peer credentials. Tomorrow open the extracted folder, run python test_swarm.py, then python swarm.py. Save one improvement and its reason in your own work journal. Record the restored backup folder and a retained run ID in your evidence; terminal history alone is not a backup.", "explanation": "A tested restore retaining runs and failures. Expected evidence: A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded.", "estimated_time": "3-8 min", "action_type": "command", "commands": { "windows": "python inspect_memory.py", "macos": "python3 inspect_memory.py", "linux": "python3 inspect_memory.py" }, "command_type": "shell", "required_inputs": [], "expected_result": "A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded.", "resources": [ "README.md", "evidence.md", "workspace-guide.md", "safety.md" ], "checkpoint": { "what_is_being_checked": "Build Bot Better and run later", "success_condition": "A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded.", "how_to_verify": "Record one recognized evidence group: run, restored, backup. Inspection notes must state the named facts. DEFERRED evidence names its reason and earns no live-success claim. Include run_id: followed by the actual 32 hexadecimal characters from the saved report.", "criteria": [ "A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded.", "I recorded actual results and labelled any deferred live check." ], "tokens": [], "user_visible_success_message": "Evidence saved: A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded.", "failure_message": "Compare the observed output with the required result before continuing.", "troubleshooting_path": "Open the help for this step.", "checkpoint_id": "L13-S02-CHECK", "tokens_any": [ [ "run", "restored", "backup" ] ], "deferred_min_reason_chars": 24, "run_id_format": true }, "troubleshooting": [ { "problem": "Observed result differs from the checkpoint", "likely_cause": "Folder, mode, permission, data shape or expected failure drill", "diagnostic_question": "At Build Bot Better and run later, what exact command or inspection failed, and how did its output differ from A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded.?", "fix": "For this step run python inspect_memory.py from the extracted kit folder. Compare against: A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "alternative_fix": "Inspect the matching downloadable guide and original conversation before retrying.", "when_to_escalate": "Stop when access, spending or data ownership is unresolved.", "bot_coach_response": "Confirm the extracted working folder, mode and exact observed state. For live failures check identity, scoped grants, warehouse and space IDs. Keep failed evidence and never substitute practice output for a live success." } ], "faqs": [ { "question": "What makes a memory backup useful?", "answer": "A tested restore retaining runs and failures. A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded." }, { "question": "Does this prove live autonomous Databricks operation?", "answer": "Only actual labelled live evidence proves the tested live action. The required lab is deterministic practice; general planning and Delta runtime integration remain extensions." }, { "question": "Why did missing config.json or a literal RUN_ID get REJECTED?", "answer": "For live mode copy config.example.json to config.json and supply your owner-approved real space IDs and authenticated identity. For local approval replace RUN_ID and SHA256 with the retained verified report path and its printed REPORT_SHA256. REJECTED exits with code 1 and is not a successful live query." }, { "question": "Python/python3 not found, wrong folder or localhost port in use", "answer": "Use Python 3.10+ in the extracted kit folder. Windows can use py -3; macOS/Linux normally use python3. For Streamlit keep the server running on your own loopback port 8082 or 8092 and open its printed address. Never paste credentials." } ], "quiz": { "question": "What makes a memory backup useful?", "choices": [ "A backup filename alone.", "A tested restore retaining runs and failures." ], "correct_answer": 1, "explanation": "A tested restore retaining runs and failures." }, "coach": { "intro": "Next: Build Bot Better and run later", "demo": "Read safety.md. Save reports, config without tokens and a copy of memory.sqlite while the process is stopped; restore to a new folder and inspect. Onboard each real worker under its own login, folder and grants; never borrow peer credentials. Tomorrow open the extracted folder, run python test_swarm.py, then python swarm.py. Save one improvement and its reason in your own work journal. Record the restored backup folder and a retained run ID in your evidence; terminal history alone is not a backup.", "handoff": "Try the action and save what actually happens.", "success": "A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded.", "failure": "For this step run python inspect_memory.py from the extracted kit folder. Compare against: A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. ", "help": "For this step run python inspect_memory.py from the extracted kit folder. Compare against: A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded. A missing script indicates the wrong working folder or incomplete extraction; retain the exact failing output before retrying. " }, "video": { "purpose": "Build Bot Better and run later", "duration_target": "45-90 seconds", "visual_type": "Narrated action outline; rendered video not supplied", "narration_script": "Graduation requires a new mission and honest scope. Durable memory, owner identity and reproducible failure evidence matter more than a self-awarded grade. Read safety.md. Save reports, config without tokens and a copy of memory.sqlite while the process is stopped; restore to a new folder and inspect. Onboard each real worker under its own login, folder and grants; never borrow peer credentials. Tomorrow open the extracted folder, run python test_swarm.py, then python swarm.py. Save one improvement and its reason in your own work journal. Record the restored backup folder and a retained run ID in your evidence; terminal history alone is not a backup. Check: A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded.", "on_screen_text": [ "Build Bot Better and run later", "A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded." ], "screen_recording_instructions": [ "Read safety.md. Save reports, config without tokens and a copy of memory.sqlite while the process is stopped; restore to a new folder and inspect. Onboard each real worker under its own login, folder and grants; never borrow peer credentials. Tomorrow open the extracted folder, run python test_swarm.py, then python swarm.py. Save one improvement and its reason in your own work journal. Record the restored backup folder and a retained run ID in your evidence; terminal history alone is not a backup.", "Show the actual result and its separate practice/live scope." ], "animation_instructions": "Highlight the action, then the corresponding result; avoid motion during command copying.", "cursor_instructions": "Point to the exact input and checkpoint; do not hide errors.", "important_visuals": [ "A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded." ], "bot_position": "right sidebar", "bot_expression": "calm", "transition": "Continue after checkpoint", "captions": [ "Graduation requires a new mission and honest scope. Durable memory, owner identity and reproducible failure evidence matter more than a self-awarded grade.", "Read safety.md. Save reports, config without tokens and a copy of memory.sqlite while the process is stopped; restore to a new folder and inspect. Onboard each real worker under its own login, folder and grants; never borrow peer credentials. Tomorrow open the extracted folder, run python test_swarm.py, then python swarm.py. Save one improvement and its reason in your own work journal. Record the restored backup folder and a retained run ID in your evidence; terminal history alone is not a backup.", "A restored memory copy retains run IDs and failures; the run-later procedure and owner boundaries are recorded." ] }, "step_id": "L13-S02", "lesson_id": "L13" } ], "resources": [ { "resource_id": "R1", "filename": "app.py", "title": "app.py", "type": "code", "contents": "\"\"\"Streamlit mission viewer; explicit run button and local human approval.\nVERSION HISTORY\n2026-10-13 EDT | chatgpt_codex | codex-2026-10-13-genie | created mission control; hub 2133.\n\"\"\"\nimport json\nfrom pathlib import Path\nimport streamlit as st\nfrom swarm import run, digest, approve\n\nst.title('Genie Swarm Mission Control')\nst.caption('PRACTICE / SIMULATION. No Databricks calls from this dashboard. Live reports can be inspected separately.')\nfault = st.selectbox('Failure drill', ['none','transient','timeout','bad-data'])\nif st.button('Run sample mission'):\n with st.spinner('Running three specialists and verifier'):\n st.session_state['report'] = run('runs', fault=fault)\nif 'report' in st.session_state:\n r = st.session_state['report']; st.metric('Mission state', r['state']); st.metric('Practice cost USD', 0)\n st.json(r); st.code(digest(r))\n reviewed = st.checkbox('I reviewed the sample totals, provenance and exact report bytes')\n if reviewed and r['state'] == 'AWAITING_HUMAN' and st.button('Approve local report'):\n st.json(approve(Path('runs')/(r['run_id']+'.json'), digest(r)))\nst.subheader('Retained mission history')\nfor path in sorted(Path('runs').glob('*.json')) if Path('runs').exists() else []:\n if path.name.endswith('.approval.json'): continue\n with st.expander(path.stem): st.json(json.loads(path.read_text()))\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R2", "filename": "app.yaml", "title": "app.yaml", "type": "worksheet", "contents": "# VERSION HISTORY\n# 2026-10-13 EDT | chatgpt_codex | Codex | codex-2026-10-13-genie | created independent course; hub 2133. Revision 2026-10-14 EDT | chatgpt_codex | codex-20261014-overnight | r7 checkpoint, metadata and expected-error repairs.\ncommand: [\"streamlit\", \"run\", \"app.py\", \"--server.address\", \"0.0.0.0\"]\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R3", "filename": "capstone.md", "title": "capstone.md", "type": "worksheet", "contents": "\n# Executive M&A due diligence: evidence gate\n\nPractice rehearsal: run the three-domain sample, inject a transient failure and a wrong metric, inspect retained history, compare independent totals and approve only unchanged passing bytes. This proves coordination machinery over invented fixtures.\n\nReal graduation: choose a new owner-approved company dataset and a fresh question spanning finance, retention and operations. Record period/currency/population, lawful access, metric definitions and independently executed reference SQL. Adapt QUESTIONS and EXPECTED deliberately; sample oracle values are invalid for a new dataset. Observe actual three-space calls, concurrency or quota-related serial limits, data provenance, incomplete/empty/truncated results, failure diagnosis, bounded retries, actual cost and an identified human review. Save report hashes and decisions. A general autonomous mission planner, distributed Delta queue, live dashboard monitoring and hard cloud budget controls remain extensions. Do not award arbitrary M&A valuation graduation from this fixed recipe.\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R4", "filename": "config.example.json", "title": "config.example.json", "type": "worksheet", "contents": "{\n \"spaces\": {\n \"finance\": \"YOUR_FINANCE_SPACE_ID\",\n \"customers\": \"YOUR_CUSTOMER_SPACE_ID\",\n \"operations\": \"YOUR_OPERATIONS_SPACE_ID\"\n }\n}", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R5", "filename": "current-route.md", "title": "current-route.md", "type": "worksheet", "contents": "\n# Current product route, checked 2026-10-13\n\nThe shared brief predates the October 12 documentation update. Supervisor Agent and the old multi-agent Apps agent-server page are now marked legacy. Follow their linked current Genie Agents and Agent Bricks CLI routes for new production systems. This course ships a custom fixed analytical SDK exercise rather than falsely claiming a fully general AI planner. A model-driven planner extension must output a validated domain/task graph, accept only allowlisted read tools, require independent data checks and retain actual live mission evidence. Live API authentication and cloud deployment are not verified by local fixture tests.\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R6", "filename": "deployment.md", "title": "deployment.md", "type": "worksheet", "contents": "\n# Optional Databricks Apps deployment\n\nTest the PRACTICE dashboard locally using loopback 8082 (8092 if busy). app.yaml listens on the platform interface only inside the hosted Databricks App; never use its 0.0.0.0 setting for a local preview. Use the workspace Apps creation/deployment UI and its current instructions to select an approved source folder containing app.py, swarm.py, requirements.txt and app.yaml. Review account authorization, cost and sharing before deployment. This dashboard intentionally runs practice only. Its service principal does not get automatic Unity Catalog privileges. Live wiring requires explicit space/data/warehouse grants under the app identity and a reviewed endpoint; local config credentials must never be uploaded. Local files are not guaranteed durable across cloud restarts: export artifacts to an approved persistent Volume or Delta implementation before real use. The supplied code does not implement that storage adapter. Hosted authentication, resource permissions, persistence, live status polling and billing meters are unverified.\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R7", "filename": "evidence.md", "title": "evidence.md", "type": "worksheet", "contents": "\n# Your evidence\n\nTrack: PRACTICE / LIVE\nIdentity (no credentials):\nDataset and metric definitions:\nMission and limits:\nRun IDs and actual commands:\nObserved overlap:\nFailure and diagnosis:\nIndependent query/oracle:\nVerified claims and remaining uncertainty:\nReport hash and human decision:\nLive cost: UNKNOWN until observed\nDeferred workspace/deployment checks:\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R8", "filename": "inspect_memory.py", "title": "inspect_memory.py", "type": "code", "contents": "\"\"\"VERSION HISTORY\n2026-10-13 EDT | chatgpt_codex | Codex | codex-2026-10-13-genie | created independent course; hub 2133. Revision 2026-10-14 EDT | chatgpt_codex | codex-20261014-overnight | r7 checkpoint, metadata and expected-error repairs.\n2026-10-14 EDT | chatgpt_codex | readable ISO timestamps and bounded run filter.\n\"\"\"\nimport argparse,sqlite3\nfrom datetime import datetime\nfrom pathlib import Path\nparser=argparse.ArgumentParser(description=\"Read retained local mission transitions; no cloud access\")\nparser.add_argument('--run-id');parser.add_argument('--limit',type=int,default=100)\na=parser.parse_args()\nif not 1<=a.limit<=10000:parser.error('limit must be between 1 and 10000')\np=Path('runs/memory.sqlite')\nif not p.exists():raise SystemExit('Run python swarm.py first')\nwith sqlite3.connect(p) as c:\n query='SELECT run_id,role,state,at,attempt FROM events'+(' WHERE run_id=?' if a.run_id else '')+' ORDER BY at DESC LIMIT ?'\n args=(a.run_id,a.limit) if a.run_id else (a.limit,)\n rows=list(c.execute(query,args))\n for run_id,role,state,at,attempt in reversed(rows):print(datetime.fromtimestamp(at).astimezone().isoformat(timespec='seconds'),run_id,role,state,'attempt',attempt)\nprint('EVENTS SHOWN',len(rows),'Use --run-id ID and --limit N to inspect another bounded slice.')\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R9", "filename": "memory-guide.md", "title": "memory-guide.md", "type": "worksheet", "contents": "\n# Memory boundary\n\npython inspect_memory.py prints local timestamped event history. SQLite stores events; JSON stores report content. Stop the runtime before copying memory.sqlite and the reports. Preserve backups in a new folder and compare run IDs before reuse. Delta templates create storage only. To build a real distributed queue add atomic conditional claims, lease expiry, idempotency keys, optimistic conflict handling, abandoned RUNNING recovery and audit correlation; verify two competing dispatchers cannot claim the same task. None is implied by ACID tables.\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R10", "filename": "queue.sql", "title": "queue.sql", "type": "worksheet", "contents": "-- VERSION HISTORY\n-- 2026-10-13 EDT | chatgpt_codex | Codex | codex-2026-10-13-genie | created independent course; hub 2133. Revision 2026-10-14 EDT | chatgpt_codex | codex-20261014-overnight | r7 checkpoint, metadata and expected-error repairs.\n-- TEMPLATE ONLY: replace YOUR_CATALOG.YOUR_SCHEMA with your approved sandbox.\n-- An append-only journal is not an atomic work claim protocol. Runtime Delta adapter not implemented.\nCREATE TABLE IF NOT EXISTS YOUR_CATALOG.YOUR_SCHEMA.swarm_missions (run_id STRING, mission STRING, mode STRING, created_at TIMESTAMP) USING DELTA COMMENT 'Codex Genie lab mission records; owner must implement runtime adapter';\nCREATE TABLE IF NOT EXISTS YOUR_CATALOG.YOUR_SCHEMA.swarm_events (run_id STRING, role STRING, state STRING, attempt INT, at TIMESTAMP) USING DELTA COMMENT 'Append-only telemetry; leases, deduplication and retry recovery require an adapter';\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R11", "filename": "README.md", "title": "README.md", "type": "worksheet", "contents": "\n# Start here\n\nExtract into a new practice folder. Python 3.10+ and standard library suffice for python swarm.py and python test_swarm.py. On macOS/Linux use python3. No account, network or paid service is required for those commands. Streamlit and live SDK are optional dependencies. Run all commands from this folder. PRACTICE uses invented fixture results, not a local AI model. Questions and planner are a fixed sample recipe. SQLite memory is not Delta integration. Keep failures and reports. Live account execution is unverified by the author.\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R12", "filename": "requirements.txt", "title": "requirements.txt", "type": "worksheet", "contents": "# VERSION HISTORY\n# 2026-10-13 EDT | chatgpt_codex | Codex | codex-2026-10-13-genie | created independent course; hub 2133. Revision 2026-10-14 EDT | chatgpt_codex | codex-20261014-overnight | r7 checkpoint, metadata and expected-error repairs.\ndatabricks-sdk>=0.60,<1\nstreamlit>=1.40,<2\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R13", "filename": "safety.md", "title": "safety.md", "type": "worksheet", "contents": "\n# Build Bot Better\n\nUse your own identity, credential variables, folder and schema; onboard workers with a named owner and minimum grants. Record changes and reasons in DEV_LOG and append script version history. Never edit another agent's work without authorization. Never remove user folders. Command-text delete guards cannot reliably detect paths computed inside scripts: inspect resolved targets, keep independent backups and test restoration. Avoid destructive cleanup commands. Treat table text as data, not executable agent instructions. Keep credentials and private query results out of shared reports. Read-only analytical specialists cannot approve external writes.\n\nDedicated local ports: core portal 8080; claude_cli 8081 (8090/8091); chatgpt_codex 8082 (8092); grok_bot 8083 (8093); grok_build 8084 (8094); antigravity_cli 8085 (8095); Ollama 11434. Bind only your allocated loopback address. If your port is occupied, use your own auxiliary port or an ephemeral loopback port.\n\nRun later: open extracted folder, inspect changes, python test_swarm.py, python swarm.py, review exact report, approve locally only when verified. Keep failed experiments and a restorable memory backup. Practice money is zero; live cloud charges remain unknown until billing evidence.\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R14", "filename": "sample.csv", "title": "sample.csv", "type": "worksheet", "contents": "region,revenue,churned,late_orders\nEast,1200,1,3\nWest,1800,2,1\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R15", "filename": "sample.sql", "title": "sample.sql", "type": "worksheet", "contents": "-- VERSION HISTORY\n-- 2026-10-13 EDT | chatgpt_codex | Codex | codex-2026-10-13-genie | created independent course; hub 2133. Revision 2026-10-14 EDT | chatgpt_codex | codex-20261014-overnight | r7 checkpoint, metadata and expected-error repairs.\n-- Use an approved sandbox catalog/schema first. Creates a new fixture; never replaces an existing table.\nCREATE TABLE bbb_genie_sample USING DELTA AS SELECT * FROM VALUES ('East',1200,1,3),('West',1800,2,1) AS t(region,revenue,churned,late_orders);\nSELECT count(*) AS rows, sum(revenue) AS revenue, sum(churned) AS churned, sum(late_orders) AS late_orders FROM bbb_genie_sample;\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R16", "filename": "swarm.py", "title": "swarm.py", "type": "code", "contents": "\"\"\"Small, inspectable analytical mission runtime. PRACTICE is deterministic, not AI.\nVERSION HISTORY\n2026-10-13 EDT | chatgpt_codex | codex-2026-10-13-genie | created bounded practice and live Genie workflow; hub 2133.\n2026-10-14 EDT | chatgpt_codex | concise REJECTED for missing configuration and literal approval placeholders; hub2136.\n\"\"\"\nimport argparse\nimport csv\nimport concurrent.futures\nimport hashlib\nimport json\nimport logging\nimport math\nimport sqlite3\nimport time\nimport uuid\nfrom pathlib import Path\n\nQUESTIONS = {\n 'finance': 'Return one row with SUM(revenue) AS revenue for all rows in the supplied sample table.',\n 'customers': 'Return one row with SUM(churned) AS churned for all rows in the supplied sample table.',\n 'operations': 'Return one row with SUM(late_orders) AS late_orders for all rows in the supplied sample table.'}\nEXPECTED = {'finance': ('revenue', 3000), 'customers': ('churned', 3), 'operations': ('late_orders', 4)}\n\ndef digest(report):\n return hashlib.sha256(json.dumps(report, sort_keys=True).encode()).hexdigest()\n\ndef verify(role, result):\n key, expected = EXPECTED[role]\n rows = result.get('rows', [])\n if len(rows) != 1 or set(rows[0]) != {key}:\n raise ValueError('Expected exactly one row and the agreed metric column')\n value = rows[0][key]\n if isinstance(value, bool) or not isinstance(value, (float, int)) or not math.isfinite(value) or value != expected:\n raise ValueError('Metric disagrees with independent sample-data oracle')\n if not result.get('sql') or not result.get('source'):\n raise ValueError('Missing query or provenance')\n return result\n\nclass Practice:\n def query(self, role, attempt, fault):\n time.sleep(.08)\n if fault == 'timeout' and role == 'finance':\n raise TimeoutError('Injected PRACTICE timeout')\n if fault == 'transient' and role == 'finance' and attempt == 1:\n raise ConnectionError('Injected PRACTICE connection failure')\n key, _ = EXPECTED[role]\n with Path(__file__).with_name('sample.csv').open(encoding='utf-8', newline='') as f:\n value = sum(int(row[key]) for row in csv.DictReader(f))\n if fault == 'bad-data' and role == 'customers': value = 999\n return {'rows': [{key: value}], 'sql': 'SELECT SUM('+key+') AS '+key+' FROM sample', 'source': 'Invented PRACTICE fixture; sample.csv'}\n\nclass Live:\n def __init__(self, config, seconds):\n from databricks.sdk import WorkspaceClient\n self.client = WorkspaceClient()\n self.spaces = config['spaces']\n self.seconds = seconds\n\n def query(self, role, attempt, fault):\n from datetime import timedelta\n # The SDK waiter polls get_message until completion; its timeout does not cancel server execution.\n space = self.spaces[role]\n m = self.client.genie.start_conversation_and_wait(space_id=space, content=QUESTIONS[role], timeout=timedelta(seconds=self.seconds))\n attachments = [a for a in (m.attachments or []) if a.query]\n if len(attachments) != 1: raise ValueError('Expected one query attachment; inspect conversation manually')\n a = attachments[0]\n raw = self.client.genie.get_message_attachment_query_result(space_id=space, conversation_id=m.conversation_id, message_id=m.id, attachment_id=a.attachment_id).as_dict()\n statement = raw.get('statement_response', {})\n if statement.get('status', {}).get('state') != 'SUCCEEDED': raise ValueError('Query result not successful')\n if statement.get('manifest', {}).get('truncated'): raise ValueError('Truncated result refused')\n columns = [c['name'] for c in statement.get('manifest', {}).get('schema', {}).get('columns', [])]\n data = statement.get('result', {})\n if data.get('next_chunk_index') is not None: raise ValueError('Multi-chunk result refused in small lab')\n rows = [{k: float(v) for k, v in zip(columns, row)} for row in data.get('data_array', [])]\n return {'rows': rows, 'sql': a.query.query, 'source': {'space': space, 'conversation': m.conversation_id, 'message': m.id, 'attachment': a.attachment_id}}\n\ndef run(folder, mode='practice', fault='none', config=None, max_calls=6, deadline=10):\n folder = Path(folder); folder.mkdir(parents=True, exist_ok=True)\n run_id = uuid.uuid4().hex\n db = folder/'memory.sqlite'\n with sqlite3.connect(db) as c:\n c.execute('CREATE TABLE IF NOT EXISTS events (run_id TEXT, role TEXT, state TEXT, at REAL, attempt INTEGER)')\n start = time.monotonic()\n backend = Practice() if mode == 'practice' else Live(config, deadline)\n import threading\n lock = threading.Lock(); calls = [0]\n def event(role, state, attempt):\n with sqlite3.connect(db, timeout=10) as c:\n c.execute('INSERT INTO events VALUES (?,?,?,?,?)', (run_id, role, state, time.time(), attempt))\n def worker(role):\n event(role, 'QUEUED', 0)\n for attempt in range(1, 3):\n with lock:\n if calls[0] >= max_calls or time.monotonic()-start >= deadline:\n event(role, 'FAILED', attempt); return {'role': role, 'state': 'FAILED', 'reason': 'Call/runtime boundary reached'}\n calls[0] += 1\n event(role, 'ASSIGNED', attempt); event(role, 'RUNNING', attempt)\n try:\n result = backend.query(role, attempt, fault)\n if time.monotonic()-start >= deadline: raise TimeoutError('Mission deadline exceeded')\n event(role, 'VERIFYING', attempt); verify(role, result)\n event(role, 'COMPLETED', attempt)\n return {'role': role, 'state': 'COMPLETED', 'attempt': attempt, 'result': result}\n except (ConnectionError, TimeoutError):\n # No automatic live resubmission: a timed-out server query can still incur charges.\n if mode == 'practice' and attempt < 2:\n event(role, 'RETRYING', attempt); time.sleep(.05); continue\n event(role, 'FAILED', attempt); return {'role': role, 'state': 'FAILED', 'reason': 'Transport/timeout failure; inspect original request before retrying'}\n except Exception:\n logging.exception('Specialist failure; review local restricted log')\n event(role, 'FAILED', attempt); return {'role': role, 'state': 'FAILED', 'reason': 'Access, schema or verification failure; no blind retry'}\n with concurrent.futures.ThreadPoolExecutor(max_workers=3) as pool:\n results = list(pool.map(worker, QUESTIONS))\n passed = all(r['state'] == 'COMPLETED' for r in results)\n report = {'run_id': run_id, 'mode': mode.upper(), 'mission': 'Revenue leakage and churn sample analysis', 'planner': 'Fixed three-domain recipe; arbitrary mission planning unsupported', 'state': 'AWAITING_HUMAN' if passed else 'FAILED', 'results': results, 'calls': calls[0], 'seconds': time.monotonic()-start, 'cost_usd': 0 if mode == 'practice' else None, 'limitations': 'Client deadlines are not cloud cancellation or a hard DBU cap. No external publication.'}\n (folder/(run_id+'.json')).write_text(json.dumps(report, indent=2), encoding='utf-8')\n return report\n\ndef approve(path, expected_hash):\n report = json.loads(Path(path).read_text(encoding='utf-8'))\n if report['state'] != 'AWAITING_HUMAN' or digest(report) != expected_hash: raise ValueError('Unverified or changed report cannot be approved')\n for row in report['results']: verify(row['role'], row['result'])\n receipt = {'run_id': report['run_id'], 'report_sha256': expected_hash, 'decision': 'human-approved-local-artifact', 'at': time.time()}\n Path(str(path)+'.approval.json').write_text(json.dumps(receipt, indent=2), encoding='utf-8')\n return receipt\n\nif __name__ == '__main__':\n Path('logs').mkdir(exist_ok=True)\n logging.basicConfig(filename='logs/swarm.log', encoding='utf-8', level=logging.INFO)\n p = argparse.ArgumentParser(); p.add_argument('--mode', choices=['practice','live'], default='practice'); p.add_argument('--fault', choices=['none','transient','timeout','bad-data'], default='none'); p.add_argument('--folder', default='runs'); p.add_argument('--config', default='config.json'); p.add_argument('--max-calls', type=int, default=6); p.add_argument('--deadline', type=float, default=10); p.add_argument('--approve'); p.add_argument('--hash')\n a = p.parse_args()\n try:\n if a.approve:\n if 'RUN_ID' in a.approve or not a.hash or a.hash=='SHA256':\n raise ValueError('Replace RUN_ID with the saved report path and SHA256 with its printed REPORT_SHA256 after review.')\n print(json.dumps(approve(a.approve, a.hash), indent=2))\n else:\n if a.mode=='live' and not Path(a.config).is_file():\n raise ValueError('Live config is missing. Copy config.example.json to config.json, enter approved real space IDs and authenticate your own Databricks identity; otherwise record DEFERRED with a reason.')\n cfg = json.loads(Path(a.config).read_text()) if a.mode == 'live' else None\n report = run(a.folder, a.mode, a.fault, cfg, a.max_calls, a.deadline)\n print(json.dumps(report, indent=2)); print('REPORT_SHA256', digest(report))\n raise SystemExit(0 if report['state'] == 'AWAITING_HUMAN' else 2)\n except (ValueError,FileNotFoundError) as exc:\n logging.exception('Input rejected')\n print('REJECTED: '+str(exc))\n raise SystemExit(1)\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R17", "filename": "test_swarm.py", "title": "test_swarm.py", "type": "code", "contents": "\"\"\"Failure, concurrency, retained history and tamper tests for the extracted lab.\nVERSION HISTORY\n2026-10-13 EDT | chatgpt_codex | codex-2026-10-13-genie | created acceptance checks; hub 2133.\n\"\"\"\nimport json\nimport logging\nimport sqlite3\nimport tempfile\nfrom pathlib import Path\nfrom swarm import run, digest, approve\n\ndef main():\n folder = Path(tempfile.mkdtemp(prefix='bbb-genie-'))\n logging.basicConfig(filename=folder/'acceptance.log', encoding='utf-8', level=logging.INFO)\n good = run(folder); assert good['state'] == 'AWAITING_HUMAN'\n with sqlite3.connect(folder/'memory.sqlite') as c:\n events = c.execute('SELECT role,state,at FROM events WHERE run_id=?', (good['run_id'],)).fetchall()\n starts = [e[2] for e in events if e[1] == 'RUNNING']\n ends = [e[2] for e in events if e[1] == 'COMPLETED']\n assert max(starts) < min(ends), 'Actual three-worker overlap required'\n retry = run(folder, fault='transient'); assert retry['calls'] == 4 and retry['state'] == 'AWAITING_HUMAN'\n assert run(folder, fault='timeout')['state'] == 'FAILED'\n assert run(folder, fault='bad-data')['state'] == 'FAILED'\n assert run(folder, max_calls=1)['state'] == 'FAILED'\n assert run(folder, deadline=.001)['state'] == 'FAILED'\n file = folder/(good['run_id']+'.json'); original_hash = digest(good)\n approve(file, original_hash)\n good['results'][0]['result']['rows'][0]['revenue'] = 123\n file.write_text(json.dumps(good))\n try: approve(file, original_hash)\n except ValueError: pass\n else: raise AssertionError('Tampered report approved')\n assert len(list(folder.glob('*.json'))) >= 7\n print('PASS: parallel overlap, bounded retry, failures, call/runtime stops, tamper rejection, retained missions', folder)\n\nif __name__ == '__main__': main()\n", "version": "1.0", "generated_from": "Build Specification resource manifest" }, { "resource_id": "R18", "filename": "workspace-guide.md", "title": "workspace-guide.md", "type": "worksheet", "contents": "\n# Live workspace setup\n\n1. Obtain approved workspace access, sandbox catalog/schema, Genie permission and SQL warehouse usage. Data privileges include USE CATALOG, USE SCHEMA and SELECT on the sample table; grant no production writes to the analytical identity.\n2. Run sample.sql once in the sandbox. Existing table name means inspect it, not overwrite it. Expected row count 2 and totals 3000/3/4.\n3. In Genie create three agents/spaces, choose that warehouse and sample table. Finance description: aggregate sample revenue; customers: aggregate churned counts; operations: aggregate late orders. Define revenue as sample currency amount, churned and late_orders as counts for the two supplied rows. No time column means no unsupported period filtering.\n4. Ask the three exact QUESTIONS from swarm.py. Inspect SQL and table results directly. Save space IDs in config.json, never tokens.\n5. Install current Databricks CLI using the official instructions, then databricks auth login --host https://YOUR_WORKSPACE_HOST using OAuth. Check databricks auth describe locally. WorkspaceClient uses unified authentication; the SDK does not create an account or grant data permissions.\n6. Install requirements into a virtual environment. Run python swarm.py --mode live --deadline 120 only when authorized. SDK/schema differences fail closed; inspect the official SDK and actual result shape before adapting. The author has not run this in a workspace.\n7. Errors: 401 means reauthenticate the correct identity; 403 means request scoped grants; missing attachment means inspect whether Genie asked clarification; 429 means stop and respect quota/retry guidance; timeout means inspect the original server conversation before resubmitting. Do not paste private results or signed download links into shared logs.\n", "version": "1.0", "generated_from": "Build Specification resource manifest" } ], "bot_coach": { "bot_name": "Buzz", "bot_role": "Offline reviewed step coach", "personality": "patient and concrete", "tone": "plain", "teaching_style": "hint, diagnostic, action, checkpoint", "introduction": "Ask about the current step. I use reviewed local answers and queue unknown questions for review.", "encouragement_style": "Focus on observed results.", "troubleshooting_style": "Ask for the smallest reproducible evidence.", "limits": "This offline coach does not call an LLM or inspect your computer. Unknown questions enter a local exportable review queue, not a shared SQL FAQ." }, "coach_interaction_model": { "stuck": "current diagnostic and first repair", "meaning": "current explanation", "error": "current troubleshooting", "why": "purpose", "skip": "require equivalent evidence", "alternative": "documented alternative", "finished": "record actual output", "start_over": "export then reset only this Build", "progressive_assistance": [ "hint", "diagnostic", "walkthrough", "resource solution" ] }, "website_specification": { "layout": "Lesson outline, actionable step, current-context coach", "features": [ "platform selectors", "command copy with selection fallback", "evidence checklist", "practical quiz", "downloadable resource ZIP", "local progress", "evidence export", "unknown-question review queue" ], "checkpoint_model": "Learner supplies observations/output; tokens check expected shape only, not host truth.", "navigation": "Browse freely; checked next-button progression retains evidence.", "video": "spec-driven walkthrough and optional speech; full rendered media is a downstream production task." }, "final_test": { "test_name": "Executive due diligence evidence gate", "instructions": "# Executive M&A due diligence: evidence gate\n\nPractice rehearsal: run the three-domain sample, inject a transient failure and a wrong metric, inspect retained history, compare independent totals and approve only unchanged passing bytes. This proves coordination machinery over invented fixtures.\n\nReal graduation: choose a new owner-approved company dataset and a fresh question spanning finance, retention and operations. Record period/currency/population, lawful access, metric definitions and independently executed reference SQL. Adapt QUESTIONS and EXPECTED deliberately; sample oracle values are invalid for a new dataset. Observe actual three-space calls, concurrency or quota-related serial limits, data provenance, incomplete/empty/truncated results, failure diagnosis, bounded retries, actual cost and an identified human review. Save report hashes and decisions. A general autonomous mission planner, distributed Delta queue, live dashboard monitoring and hard cloud budget controls remain extensions. Do not award arbitrary M&A valuation graduation from this fixed recipe.\n", "required_inputs": [ "capstone.md", "evidence.md" ], "success_conditions": [ "Define three scoped analytical work orders.", "Run concurrent lanes and retain state history.", "Reject invalid data and bound retry attempts.", "Bind approval to verified report bytes.", "Prepare authenticated Genie specialist queries without borrowing credentials.", "Distinguish observed practice results from unverified cloud and general autonomy claims." ], "automated_tests": "python test_swarm.py", "manual_tests": "Actual live query, deployment, business verification and approval evidence if live graduation is claimed.", "expected_output": "Retained verified practice report and hash receipt; live graduation separately evidenced.", "failure_paths": "Use the specific step diagnostic, retain failure evidence and stop when access is required." }, "extensions": [ "Implement a tested Delta queue adapter with leases and idempotency.", "Use current Agent Bricks tooling for model-driven mission planning with validated tools.", "Add durable Apps storage, authenticated approvals and observed billing telemetry." ], "metadata": { "SEO_title": "Build Your Own AI Swarm in Databricks with Genie | BuildBotsBetter", "SEO_description": "Create a working free PRACTICE swarm and prepare a scoped live Genie SDK workflow with persistent evidence, failure drills and human approval.", "SEO_keywords": [ "Databricks", "Genie", "AI swarm", "parallel analytics" ], "difficulty": "beginner", "related_builds": [ "dogfooding", "buildbotsbetter-onboarding" ], "next_recommended_build": "dogfooding" }, "automation_map": [ { "output": "website", "generated_from": "BUILD_SPEC.json", "human_review_required": true }, { "output": "walkthrough", "generated_from": "BUILD_SPEC.json", "human_review_required": true }, { "output": "bot dialogue", "generated_from": "BUILD_SPEC.json", "human_review_required": true }, { "output": "resources", "generated_from": "BUILD_SPEC.json", "human_review_required": true }, { "output": "quiz", "generated_from": "BUILD_SPEC.json", "human_review_required": true }, { "output": "captions", "generated_from": "BUILD_SPEC.json", "human_review_required": true }, { "output": "metadata", "generated_from": "BUILD_SPEC.json", "human_review_required": true } ], "quality_assurance_report": { "scope": "Course materials and free local lab; cloud and general autonomy limits explicit", "tests": "Extracted resource acceptance, browser navigation and visual inspection", "unverified": [ "Live Databricks SDK calls, account permissions and deployment have not been executed by the author.", "Required lab uses deterministic fixture workers and a fixed planner, not autonomous general AI planning.", "SQLite runtime is not integrated with Delta; templates do not implement distributed claiming or durable Apps storage.", "Dashboard displays completed practice runs, not live progress, tokens or DBU meters.", "Local limits cannot cancel remote jobs or guarantee a hard cloud spend ceiling.", "Practice sample metrics do not establish arbitrary M&A valuation or causal business conclusions.", "macOS/Linux commands are displayed but actual execution unverified; no rendered video claimed." ], "repairs": "2026-10-14: all checkpoints require named facts/results; clear optional deferred track; distinct quizzes; failed command evidence rejected; missing config/approval placeholders return concise diagnostics." }, "sources": [ { "title": "Official Genie setup", "url": "https://docs.databricks.com/aws/en/genie-agents/set-up" }, { "title": "Official Genie Python SDK", "url": "https://databricks-sdk-py.readthedocs.io/en/latest/workspace/dashboards/genie.html" }, { "title": "Official Genie conversation API", "url": "https://docs.databricks.com/aws/en/genie-agents/conversation-api" }, { "title": "Legacy Supervisor and current route notice", "url": "https://docs.databricks.com/aws/en/agents/agent-bricks/multi-agent-supervisor" }, { "title": "Databricks Apps setup", "url": "https://docs.databricks.com/aws/en/dev-tools/databricks-apps/" }, { "title": "Shared course requirements", "url": "../../specs/databricks-genie-swarm.md" } ], "known_uncertainties": [ "Live Databricks SDK calls, account permissions and deployment have not been executed by the author.", "Required lab uses deterministic fixture workers and a fixed planner, not autonomous general AI planning.", "SQLite runtime is not integrated with Delta; templates do not implement distributed claiming or durable Apps storage.", "Dashboard displays completed practice runs, not live progress, tokens or DBU meters.", "Local limits cannot cancel remote jobs or guarantee a hard cloud spend ceiling.", "Practice sample metrics do not establish arbitrary M&A valuation or causal business conclusions.", "macOS/Linux commands are displayed but actual execution unverified; no rendered video claimed." ], "checkpoint_revision": "r13-20261017", "one_sentence_description": "Create a working free PRACTICE swarm and prepare a scoped live Genie SDK workflow with persistent evidence, failure drills and human approval.", "final_outcome": "An inspectable three-lane analytical practice swarm with retained reports, failure checks and hash-bound human review; live cloud graduation requires separate observed evidence.", "tags": [ "Databricks", "Genie", "swarm", "parallel analytics", "human approval" ], "required_hardware": [ "A computer with Python 3.10+ and a terminal for practice.", "Internet and owner-approved Databricks workspace access are optional for the separate live track." ] }