100%

End-to-end validation + final gate

Goal

Prove the deliverable on the real production path and seal the regression + autopilot + self-test gates. The binding functional acceptance test drives the fixture through the real /continue-roadmap serve loop end to end; the autopilot invariant (zero AskUserQuestion) is asserted; all runtime self-tests + plan_corpus check are green.

Implementation sketch

test_grind_escalation_e2e.py drives the real serve loop on the fixture and asserts (1) new forward section authored, (2) served next, (3) K-fruitless fall-through to mission_reground. Autopilot-pause guard surface asserts zero AskUserQuestion. Self-tests: create_plan_runtime, plan_orchestrator, ledger_evolution_runtime; plan_corpus check.

Spec references

CLAUDE.md §Fix Completeness (production-path e2e + self-tests); skill-control-contract.md §Autopilot Mode (zero AskUserQuestion); learning-ledger.md §5 (append-only); tests.md §Production-path end-to-end test.

Work Items

  • Author the binding end-to-end functional acceptance test (the plan-wide functional criterion probe): the s00 fixture driven through the REAL /continue-roadmap serve loop auto-authors a fresh forward section via create-plan-inline AND serves it next; after K fruitless fresh sections it falls through to mission_reground. The real production entry point, never a seam-injected stub (layer-coverage L12).

  • Assert the autopilot invariant end-to-end: the entire escalation path emits ZERO AskUserQuestion (grind-forever-never-ask) — the create-plan-inline section authoring auto-resolves its own grill + Step 9 per skill-control-contract §Autopilot. Verify via the autopilot-pause guard surface.

  • Green-gate: python -m scripts.create_plan_runtime --self-test, python -m scripts.plan_orchestrator --self-test, python scripts/ledger_evolution_runtime/.../--self-test (runtime self-tests), python -m scripts.plan_corpus check plans/grind-exhaustion-forward-escalation, and the full new pytest suites all pass; no append-only ledger invariant violated by the auto-authored sections.

Fresh intel (regenerated)

{ “schema_version”: 1, “target”: { “kind”: “plan-section”, “ref”: “grind-exhaustion-forward-escalation/s-20273fc6” }, “generated_at”: “2026-06-19T18:22:21.751073+00:00”, “graph_state”: { “head_sha”: “dbb10f0e”, “last_code_import_at”: “2026-06-19T17:39:41.692Z”, “embedding_stale”: false, “insights_stale”: false, “cpg_stale”: false }, “surfaces”: {}, “summary”: “intel-package[plan-section:grind-exhaustion-forward-escalation/s-20273fc6] agent-authored dossier”, “degraded”: false, “dossier”: { “objective_symbols”: [], “tiers”: {}, “agent_authored”: true, “difficulty”: { “class”: “routine”, “research_online”: false, “research_mode”: “auto”, “signals”: [], “plan_dir”: “/home/eric/projects/ori_lang/plans/grind-exhaustion-forward-escalation” } } }

(full dossier: end-to-end-validation—s-20273fc6.intel.json)