Release: Runtime And Learning Polish

Forgeflow 4.3.14 tightens the post-release loop: release readiness explains environment blockers, installed runtime verification checks every managed helper, project intelligence produces a next-work brief, living-map guidance reaches reviewers, and pilot evidence decisions now explain why to repeat, expand, stop and fix, or defer.

Release And Install Verification

• /forgeflow-release-readiness now labels blockers as command failures, allowlist issues, missing commands, missing release-check source, or execution-environment blockers.
• /forgeflow-version verifies installed runtime helpers against the managed manifest and reports missing or invalid helper slots with a repair action.
• Repair guidance falls back to the installed updater helper or a local checkout when /update-forgeflow is unavailable.

Project Intelligence

• build-project-intelligence.js now writes a next-work brief with read-first, avoid-first, validate-first, and proof-boundary guidance.
• The brief keeps lanes semantic: runnable refresh commands stay in readiness and review-prep, not in read or validation lanes.
• Context packets now include living project-map guidance and synthesis input includes structured living_map_guidance.

Pilot Evidence

• Pilot rollups aggregate project-intelligence readiness, living project-map status, and agent-feedback signal.
• Adoption decisions now include an explicit explanation across setup friction and the three state-aware signal lanes.
• State-aware signals are decision-relevant, and rollup Markdown sanitizes support categories before rendering public-safe summaries.

Validation

• Each phase passed focused tests, full scripts/forgeflow/test-*.js, docs/release guards, and Forgeflow review.
• Review findings around ENOENT coverage, runtime helper integrity, lane semantics, static-analysis boundaries, decision relevance, and support-category privacy were fixed before release.