| alwaysApply | true |
|---|
- Three layers: SoT files (primitive) → EPICs (compose SoT) → PRD stage (composes both). All scores write to
status/readiness.json. - Compute: Run
python scripts/readiness.py runto refresh. Output survives instatus/readiness.jsonwithlast_computedtimestamp. - Inspect:
readiness.py status(text report) orreadiness.py status --json(machine-readable). - Thresholds:
score ≥ 70PASS,50–69WARN,< 50BLOCK. Exit codes 0/1/2 match. - Inputs: Declared in
readiness_inputs:frontmatter — PRD.md for stage scope,epics/EPIC-XX.mdfor epic scope. Seedocs/READINESS_PROTOCOL.mdfor schema. - Dimension overrides: Use
dimension_overrides: { confidence_avg: disabled }per item when the repo hasn't adopted a convention. Disabled dimensions drop; remaining weights renormalize. - Traceability: EPIC caps cite
caused_bySoT file; SoT blocks listconsumed_by_epics. Agents follow the causal chain to find root-cause leverage. - Before advancing gates: Run readiness. If
summary.current_stage.score < threshold_warn, update the EPIC and STOP (reinforces rule 05).