Skip to content

Latest commit

 

History

History
62 lines (46 loc) · 1.99 KB

File metadata and controls

62 lines (46 loc) · 1.99 KB

Workflow Modes

Detailed workflow preset comparison moved from README.md.

Presets

Preset triggerMode captureMode autoApproveSafeCandidates
ultra-safe-manual manual direct false
recommended-semi-auto detect candidate false
automation-first detect candidate true

Axis Definitions

  • triggerMode
    • manual: extraction only via explicit CLI calls
    • detect: hooks and brain capture auto-detect extraction opportunities
  • captureMode
    • direct: accepted memory becomes active immediately
    • candidate: new memory starts reviewable as candidate
    • reviewable: candidate-like mode that also defers merge/supersede decisions
  • autoApproveSafeCandidates
    • enables strict safe auto-promotion path when true

Recommended Default Loop

recommended-semi-auto is the default for most repositories:

  1. first conversation in session: brain start (or brain inject as fallback)
  2. fresh conversation later in same session: brain conversation-start
  3. session end: extract to candidate queue
  4. review: brain review
  5. quick pass: brain approve --safe
  6. edge cases: brain approve <id>
  7. hygiene: brain score and brain sweep --dry-run

Safe Auto-Approve Rules

Auto-promotion applies only when all checks pass:

  • reviewer decision is accept with reason novel_memory
  • type is not working
  • content is not temporary/noise
  • no merge/supersede/reject signals

Otherwise, the item remains for manual review.

Mode Selection Guidance

  • Choose ultra-safe-manual for strict human-controlled repos
  • Choose recommended-semi-auto for balanced everyday usage
  • Choose automation-first only when team quality gates and review discipline are stable

Useful Commands

brain init --workflow recommended-semi-auto
brain setup --workflow recommended-semi-auto
brain setup --workflow ultra-safe-manual
brain setup --workflow automation-first
brain promote-candidates --dry-run