fix(ai): fail loud on stale handoff PR state (#13985)

[neo-gpt]

Resolves #13985

Golden Path now fails loud when the synthesizer does not prove a handoff write: golden-path records a failed task outcome with reasonCode: golden-path-handoff-write-unverified unless wroteHandoff: true is returned. The Active PR Cycle renderer is centralized, emits a freshness expiry for live PR snapshots, and replaces stale PR blocks with an explicit degraded section when GitHub PR fetch fails instead of preserving old lifecycle data.

Evidence: L2 focused unit/static/config lint is complete. Residual: L3 live post-merge validation should confirm the next orchestrator Golden Path run updates resources/content/sandman_handoff.md and no longer leaves stale PR lifecycle state behind a completed task marker.

Deltas from ticket

• Added goldenPathActivePrStateFreshnessMs / NEO_GOLDEN_PATH_ACTIVE_PR_STATE_FRESHNESS_MS with a 60 minute default for Active PR Cycle State freshness.
• Kept repo-enrichment failure local to the handoff output: fetch failures render degraded PR state but do not block the rest of the Golden Path write.
• Added scheduler-level write-proof enforcement so a silent/undefined synthesizer return can no longer mark golden-path complete.

Config Template Change

Changed key:

• goldenPathActivePrStateFreshnessMs / NEO_GOLDEN_PATH_ACTIVE_PR_STATE_FRESHNESS_MS

Local config follow-up:

• No manual value is required; the default is active.
• After merge, run the config overlay update script so gitignored local config.mjs files understand the new shape.
• Restart the orchestrator after pulling dev; running daemons need the new code and config leaf loaded.

Test Evidence

• node --check ai/services/graph/GoldenPathSynthesizer.mjs
• node --check ai/daemons/orchestrator/scheduling/pipeline.mjs
• node --check test/playwright/unit/ai/services/graph/GoldenPathSynthesizer.spec.mjs
• node --check test/playwright/unit/ai/daemons/orchestrator/scheduling/pipeline.spec.mjs
• node --check test/playwright/unit/ai/daemons/orchestrator/Orchestrator.spec.mjs
• npm run test-unit -- test/playwright/unit/ai/daemons/orchestrator/scheduling/pipeline.spec.mjs -> 17 passed
• npm run test-unit -- test/playwright/unit/ai/daemons/orchestrator/Orchestrator.spec.mjs -> 56 passed
• npm run test-unit -- test/playwright/unit/ai/services/graph/GoldenPathSynthesizer.spec.mjs -> 36 passed
• Exact-head self-review spot checks: pipeline missing-handoff proof 1/1, Active PR Cycle State 2/2, full Orchestrator.spec.mjs 56/56.
• npm run ai:lint-config-template-ssot -> passed
• npm run agent-preflight -- ai/daemons/orchestrator/scheduling/pipeline.mjs ai/mcp/server/memory-core/config.template.mjs ai/services/graph/GoldenPathSynthesizer.mjs test/playwright/unit/ai/daemons/orchestrator/Orchestrator.spec.mjs test/playwright/unit/ai/daemons/orchestrator/scheduling/pipeline.spec.mjs test/playwright/unit/ai/services/graph/GoldenPathSynthesizer.spec.mjs -> passed
• git diff --check -> passed

Post-Merge Validation

• Pull dev, run node ./ai/scripts/setup/initServerConfigs.mjs --migrate-config, then restart the orchestrator.
• Confirm the next Golden Path run writes resources/content/sandman_handoff.md with an Active PR Cycle State freshness expiry or degraded PR-state marker.
• Confirm golden-path task state is not marked completed unless the synthesizer reports wroteHandoff: true.

Commits

• 816d146f0d - fix(ai): fail loud on stale handoff PR state (#13985)

Authored by Euclid (GPT-5, Codex Desktop). Session 019ef378-527d-7393-bc74-ec3a1d3f2ddf.