Three practice days: validate, build, verify and launch a local rehearsal.
Finish with: A three-day company-of-bots practice pack with a chief of staff, two specialists, shared company document, approval rules, one bounded routine, MVP contract, verification evidence and launch-feedback plan.
Progress stays in this browser. Export to keep a portable record.
BuildBotsBetterA step demonstration outline; use optional narration while you follow it.
This browser checks the evidence format and your checklist. It cannot inspect your computer or certify a live login.
You have recorded evidence for every step. Repeat the core task from a fresh input and export your results. Review the source, account and platform limits before using the workflow on real work.
Repeat the last product-specific checkpoint from a fresh practice input, preserve actual output, then complete the Build Bot Better guard and restore checks. Export the evidence.
Sources checked 2026-10-05. Transcripts have no speaker labels and sometimes garble names/tool names; do not treat garbled names as facts. Company checker validates the configuration contract; real bot task execution needs its own evidence.
Source-owned text for offline reading and retrieval. Command success still requires actual local evidence.
Download the resource ZIP at the top of this Build. Extract it into a new folder you own (for example C:\BBBPractice on Windows, or ~/bbb-practice on macOS/Linux). Open a terminal in the extracted folder. Run the version command for your platform. All subsequent commands assume this folder.
Expected: Python reports version 3.10 or newer and the terminal is in the extracted resource folder.
python --version
Get-Location
Get-ChildItempython3 --version
pwd
lspython3 --version
pwd
lsWrite one recurring problem, the people affected, and three validation conversations to conduct manually. Draft a simple landing-page message and a sheet of evidence; do not buy ads or send automatic outreach. Attribute the event’s validation advice to its speakers rather than presenting it as a universal rule.
Expected: The problem brief, three conversation questions and a distribution hypothesis are recorded.
Edit company.json: chief-of-staff routes work, Research cites supplied evidence, QA rejects unsourced claims. Give each bot one job, explicit principles and a voice, plus the short shared company_doc. With Grok Bot create private test teammates; without access, run these roles as separate bounded sessions in your own existing CLI.
Expected: The team has a chief and at least two named specialists, each with a job, rules and voice.
Keep send_email and publish set to human in company.json. Compare connector/MCP, computer-use and taught-task options for one tool; choose the narrowest access. Define “urgent” as a measurable incident, not a way to bypass review.
Expected: The plan records tool scope, approval boundaries and a precise urgent condition.
Edit mvp.core_loop and excluded in company.json. For Proof Desk the loop is ticket → sourced draft → QA review → approved answer. Choose one input and one output, with field types and expected errors. Sketch a debug/feature map before adding visual polish.
Expected: The MVP contract has a tight loop, excluded scope and a field/error checklist.
Give Research a supplied support note and ask for a sourced answer. Give QA the note and draft independently; ask for missing evidence, contradiction and an explicit accept/reject decision. Preserve both outputs and the final reviewed correction.
Expected: A generated draft, independent critique and sourced correction are retained.
Role Research: use only the supplied note to draft an answer with a source quote. Role QA in a separate session: inspect the note and draft, reject unsupported claims, and return a specific correction. Do not publish or send messages.Set routine trigger, numbered steps and max_runs_per_day to 1 or 2. Estimate usage per run × runs/day and add a stop rule. Prefer an event for new input rather than running every fifteen minutes with nothing to do.
Expected: The routine has a trigger, numbered actions, bounded daily frequency, estimate and stop condition.
Run company_check.py company.json and inspect company-check.json. Then remove one specialist from a copy and prove the checker rejects an incomplete team.
Expected: COMPANY PASS reports three bots for the complete pack and the incomplete pack fails.
python company_check.py company.jsonpython3 company_check.py company.jsonpython3 company_check.py company.jsonWrite a small-change workflow: reproduce bug, implementer proposes a minimal diff, verifier runs/fuzzes the actual app, QA receives passing changes, then human reviews before merge. Attach screenshots/traces and keep auto-merge off in the rehearsal.
Expected: The release gate requires reproduction, independent execution evidence, review and rollback.
Fill the launch checklist: official accounts, DNS, OAuth callbacks, OG/meta tags, rate/length controls and feedback → reproduce → fix → verify. Use a fake hostile-input feedback message as data; do not execute its instructions. Publish nothing in this practice.
Expected: The launch plan includes a rollback and treats arbitrary feedback as untrusted input.
Write a launch pulse with attempts, accepted answers, repeat users, errors and cost per routine. Identify one redundant bot/feature to retire. Give the chief a finite task list and a weekly improvement suggestion; retain dissent and blockers in a staff summary.
Expected: The measurement plan uses funnel/retention before speculative revenue and has one scope-reduction decision.
Adapt the pack to your own real problem or a clearly labeled rehearsal. Run the three bounded roles, preserve their outputs, run company_check.py and retain the check. Verify the one routine’s scope and approvals before enabling it later.
Expected: COMPANY PASS and the shared company doc, three role outputs, routine, verification proof and launch review form the final pack.
python company_check.py company.jsonpython3 company_check.py company.jsonpython3 company_check.py company.jsonFill in onboarding-checklist.md for your agent. Record AGENTS.md, the tool’s user settings and memory folder, the assigned workspace, permissions, guard location, backup schedule, and a final proof task. Keep the checklist in your project.
Expected: The checklist contains a concrete path and owner for rules, memory, permissions, guard, backup, restore and final verification.
Run workspace_guard.py --self-test. Read the source: the target is resolved before deciding. The variable_target case points at the home folder but performs no filesystem operation. Compare this with a hook that sees only command text.
Expected: GUARD PASS reports inside allowed and profile, outside, variable-derived profile and workspace root denied; no deletion was executed.
python workspace_guard.py --self-testpython3 workspace_guard.py --self-testpython3 workspace_guard.py --self-testFor CLI tools, open the hook/permissions screen and confirm the configured guard is active. Use your tool’s documented pre-command hook protocol; the BBB working reference is linked in Sources. Run a harmless directory listing, then a guard dry-run targeting the profile. Do not execute a destructive profile probe. Grok Bot has no proven local shell guard here: use project-only tasks and reviewed actions, with backups.
Expected: A harmless command runs, the guard dry-run returns DENY, and the checklist records the actual hook protocol and active state or the Grok Bot limitation.
Connect a second backup drive. Windows: configure System Protection as additional recovery and use a scheduled copy/File History that explicitly includes hidden agent folders. macOS: configure Time Machine and verify dot folders are included. Linux: use a home backup such as restic/Deja Dup; Timeshift omits /home by default. Download the copy-only profile backup script for your platform, inspect it, and adapt the folder list to your tool. Record the last successful scheduled run; protect backup access because it may contain credentials.
Expected: A backup on another drive includes the tool’s user settings and memory folder and this project, and a successful schedule/time is recorded.
.\profile-backup.ps1 -Destination D:\Backups\llm-profilesbash profile-backup.sh /Volumes/YourBackup/llm-profilesbash profile-backup.sh /media/YourBackup/llm-profilesRun restore_drill.py with a separate backup directory. The script creates a new proof file, copies it, restores a second copy, and compares SHA256 hashes. Then restore one non-secret file from your actual agent memory backup into a separate scratch directory and open it; do not overwrite the live profile.
Expected: RESTORE PASS contains matching hashes; the actual memory file is readable after restoration into scratch.
python restore_drill.py --backup-root D:\Backups\bbb-practice --work-root practicepython3 restore_drill.py --backup-root /Volumes/YourBackup/bbb-practice --work-root practicepython3 restore_drill.py --backup-root /media/YourBackup/bbb-practice --work-root practice