This protocol anchors work to canonical system memory before any changes are made.
If any of the following are true: • More than 7 days since last repository activity • Returning after travel, illness, burnout, or interruption • Uncertain about current phase, locks, or allowed work • Handing control to an assistant (Cursor or human) • Feeling disoriented about what is real vs assumed
Then STOP.
Run docs/PROTOCOL_REENTRY_AND_PROGRESS.md in full before continuing.
No code may be written until the Re-Entry Protocol is complete and
PROGRESS_REPORT.md exists for the current date.
This makes re-entry structural, not optional.
What NOT to do
Do not auto-fill intent or exit conditions. Those must remain human choices.
The system protects you from drift, not from decision-making.
Mechanical enforcement (required):
- Run
./scripts/reentry_guard.shbefore executing this protocol.
- Current branch:
- Uncommitted changes:
- Build status: ☐ Passing ☐ Failing ☐ Not evaluated (reason)
- Recent commits reviewed:
- Core memory index reviewed: docs/0_START_HERE.md
- Engine state: ☐ Open ☐ Closed ☐ Closing
- Observer layer: Not implemented (by design)
- Cockpit / UI work: ☐ Allowed ☐ Not allowed today (must choose)
Affirmations:
- No new insight types will be added today
- No new data models will be added today
- No new metaphors will be introduced today
- Work today must articulate existing vision, not expand it
- Confirm Observation Language Invariant is active (see
docs/ARCHITECTURE_NOW.md) - If doing any UI work today, run:
If it exits nonzero, classify each match as OK, Soft interpretation, or Violation and fix violations before proceeding. Log results in START_OF_DAY.
./scripts/observation_language_scan.sh
The system remembers patterns, tensions, and shifts — not goals, judgments, or prescriptions.
Describe the class of work, not the tasks. Example: "Stabilization," "Articulation," "Verification," "Cleanup."
What must be true at the end of today for the work to be considered complete?