| name | brain-schedule |
| description | Create or update the Company Brain operating schedule for human-gated or autonomous runs. Use when the user says schedule the brain check, set the morning review, automate daily checks, run without a human, keep human in the loop, set cron, or define auto-capture guardrails. |
Brain Schedule
Generate the operating schedule. The schedule is about checks, staging, review, and maintenance; it is not a data-population shortcut.
Workflow
- Resolve the brain root.
- Choose
human mode for the first two weeks unless the user explicitly asks for autonomous mode.
- Preview
schedule.md.
- Write only with explicit permission or when the user asked to update the schedule.
- Include daily checks: connections, sources, health, lint,
ADD_TO_BRAIN/ digest when present, staged queue, and punch list.
- State what can run unattended and what always needs human approval.
Command Pattern
Preview:
python3 <plugin-root>/bin/brain-schedule.py \
--root "$BRAIN_ROOT" \
--mode human \
--time 08:30 \
--timezone local \
--champion "<Brain Owner>" \
--operator "<Brain Operator>"
Write:
python3 <plugin-root>/bin/brain-schedule.py \
--root "$BRAIN_ROOT" \
--mode human \
--time 08:30 \
--timezone local \
--champion "<Brain Owner>" \
--operator "<Brain Operator>" \
--write
Modes
human: checks and staging can run, promotion requires explicit human approve/reject/revise.
autonomous: approved sources may stage proposals unattended; auto-promotion is only for explicit low-risk categories.
simple-team: not a schedule mode, but the schedule should include drop-zone digest and low-risk manual contribution handling when the profile is active.
Guardrails
- Meeting-derived, HR, legal, finance, customer-confidential, and strategy-changing notes stay human-approved.
- Personal connectors never run as scheduled company-brain sources.
- Source registry and lint must run before scheduled capture is trusted.
- The schedule should name a Brain Owner and a Brain Operator, not depend on one hidden local machine.