Operator Playbook¶
Daily¶
Morning¶
python3 scripts/validate_index.py
python3 scripts/validate_vault.py examples/fake-vault
# Optional, for a connected private vault:
# gbrain sync --repo ~/private-vault --no-pull --no-embed
# gbrain embed --stale
# gbrain health
Review:
- today's calendar
- upcoming meetings
- inbox items
- open decisions
- stale follow-ups
During the Day¶
For every important meeting:
- create or update people pages
- create meeting note
- extract open threads
- add follow-up actions
- link companies and projects
Evening¶
Process inbox/:
- promote durable facts into entity pages
- archive noise
- create decision records
- create or update project notes
Weekly¶
Run validation and retrieval checks:
python3 scripts/validate_index.py
python3 scripts/validate_vault.py examples/fake-vault
# Optional private-vault checks:
# gbrain health
# gbrain query "what strategic risks changed this week?"
# gbrain query "which relationships need follow-up?"
# gbrain query "what decisions are waiting for evidence?"
Produce:
- weekly strategic review
- relationship follow-up list
- open decision register
- stale claim list
Monthly¶
Run a maintenance review:
- duplicate entity scan
- dead links
- orphan pages
- outdated project pages
- unembedded chunks
- unsourced claims
Executive Workflows¶
Meeting Prep¶
Prompt:
Prep me for the meeting with [person/company].
Expected output:
- who they are
- relationship history
- current relevance
- risks or sensitivities
- what to ask
- what not to say
- source slugs
Relationship Map¶
Prompt:
Who do we know at [company], ranked by relationship strength?
Expected output:
- top people
- roles
- last interaction
- best intro path
- confidence level
Strategic Brief¶
Prompt:
Brief me on [topic] using the brain, not generic web knowledge.
Expected output:
- answer first
- cited evidence
- inference clearly labeled
- recommended action
Book Extraction¶
Prompt:
Turn this book into mental models for our operating system.
Expected output:
- core thesis
- 5-10 mental models
- examples
- implications
- links to existing concepts
Quality Bar¶
A good answer is:
- concise
- sourced
- current
- decision-relevant
- explicit about confidence
- clear about what is fact vs inference
A bad answer:
- sounds generic
- has no slugs
- invents relationship context
- ignores recent notes
- gives a balanced essay when a decision is needed