Enforce Cerebro-first protocol by identifying domain, reading source docs and recent logs, building Known State, and citing rules before all actions.
Before any substantive action, run this sequence:
companies/bda, companies/mission, companies/bms, companies/clawstation, personal/trading, or operations/*.COMPANY.md first (if present), then the most specific SKILL.md/runbook.memory/YYYY-MM-DD.md for today and yesterday.Do not skip this gate unless the user explicitly requests a one-shot answer unrelated to operations.
Use tag-first retrieval with the index at references/cerebro-index-v2.1.md:
#bms, #clawstation, #trading, #cron, #jira, #incident).Use references/domain-routing-v2.1.md to route tasks:
cerebro/companies/<name>/...cerebro/operations/...cerebro/personal/trading/...cerebro/runbooks/...cerebro/vendors/...Pass the "where does it say that?" test before acting.
If chat instruction conflicts with Cerebro:
Priority order: Cerebro > SOUL/AGENTS > MEMORY > chat history.
When a decision/rule/process changes:
memory/YYYY-MM-DD.md with:
Never leave process changes only in chat.
Always return a 4-line snapshot before proceeding:
If needed doc is missing:
references/missing-doc-template.md.No durable updates needed.references/startup-checklist.md — fast pre-action checklistreferences/decision-log-template.md — standard decision record formatreferences/missing-doc-template.md — template for new Cerebro docsreferences/cerebro-index-v2.1.md — lightweight tag indexreferences/domain-routing-v2.1.md — task routing matrixreferences/scenario-profiles-v2.1.md — setup profiles for solo/agency/multi-agent teamsreferences/enforcement-checklist-v2.1.md — pre-done anti-drift checksreferences/done-schema-v2.1.md — proof-based status output schemaZIP package — ready to use