Skip to content

Historical field note

This record preserves ground-truth investigation evidence for a single dream finding. It is not a current operating contract; use the numbered book and current runbooks for instructions.

Ground Truth: dream finding dreamrepair:965c6e89c762d29f07df25aafd3ac96f (openclaw_fleet_rollout deliverable)

Task: Establish the ground truth for a low-confidence, project-scope dream finding scoped to the openclaw_fleet_rollout deliverable for project mac: determine whether a live, reproducible defect exists or whether the finding is an evidence-volume artifact, and record the evidence gap. Finding: kind failure_pattern, scope project, repo_area mac, fingerprint dreamrepair:965c6e89c762d29f07df25aafd3ac96f, confidence low (score 0.35), backed by exactly one evidence record. Evidence record: mem_4ad6f3bf11fa4534a1569edec925198b (record_type deployment_learning:mac), origin task task_37e658cac7d643aebbed4742a25777a2, whose recap concerns the openclaw_fleet_rollout deliverable audit. Prepared by: fleet worker (investigation node; no production code, test, skill, or deploy edits).

Verdict: NO LIVE REPRODUCIBLE DEFECT — evidence-volume artifact, not actionable

The deliverable is present and intact, its dedicated suite is green under the canonical hermetic runner, and the single supporting record carries no failing assertion, stack trace, current reproduction, or named offending code path. The 0.35 confidence is the classifier's deterministic single-record structural floor, i.e. an evidence-volume signal, not a confirmed defect. This adopts and independently re-verifies the same ground truth reached for the near-duplicate finding dreamrepair:3dc2cf... (closed NOT ACTIONABLE in docs/archive/field-notes/closeout-dreamrepair-3dc2cf-openclaw-fleet-rollout.md).

No files under src/mac/, tests/, skills/, or deploy/ were modified.

Deliverable re-verified present and intact

src/mac/openclaw_fleet_rollout.py exposes every required public symbol, and all import cleanly against the bootstrapped .venv:

  • ROLLOUT_PLAN_SCHEMA = "mac.openclaw_fleet_rollout.v1" (src/mac/openclaw_fleet_rollout.py:22)
  • RolloutPlanStep (src/mac/openclaw_fleet_rollout.py:31)
  • RolloutPlan (src/mac/openclaw_fleet_rollout.py:41)
  • RolloutResult (src/mac/openclaw_fleet_rollout.py:65)
  • build_staged_rollout_plan (src/mac/openclaw_fleet_rollout.py:83)
  • execute_staged_rollout (src/mac/openclaw_fleet_rollout.py:139)

Canonical hermetic runner result

Command (target suite only):

scripts/run-contract-tests.sh tests/test_openclaw_fleet_rollout.py -q

Result: 52 passed (exit code 0). The suite (tests/test_openclaw_fleet_rollout.py:1) covers builder validation, canary/promote staging, the executor simulate/deploy/health paths, and a module-contract regression guard. There is no failing test, assertion, trace, or reproducer to repair.

Evidence-record analysis and the evidence gap

  • Single, non-reproducing record. The only supporting evidence (mem_4ad6f3bf11fa4534a1569edec925198b, deployment_learning:mac) is a historical deployment/learning recap from an implementation-audit task (task_37e658cac7d643aebbed4742a25777a2). It carries no failing assertion, stack trace, current reproduction, or named offending code path. The record id does not appear anywhere in the checked-out repository.
  • No independent corroboration. Nothing in the repository — neither the module, the 52-test suite, nor the hermetic runner — reproduces a rollout failure.
  • Score is a structural floor, not a severity signal. The 0.35 confidence is the classifier's deterministic single-record floor (CONFIDENCE_THRESHOLDS["low"] = ("low", 0.35), src/mac/dream_cycle_classifier.py:86), returned by _confidence_for(...) whenever evidence_count < 2 and the candidate is not high/medium (src/mac/dream_cycle_classifier.py:233). Low confidence here reflects evidence volume (one record), not defect severity.

Follow-up recommendation

  • Close the finding as NOT ACTIONABLE. No source, skill, or tool change is warranted; the deliverable and its tests are intact and green.
  • No new test/guard is added. The module-contract regression guard already present in tests/test_openclaw_fleet_rollout.py is the appropriate standing guard for this deliverable; adding another would be redundant.
  • Systemic note (no code change): single-record deployment_learning recaps of implementation/audit tasks structurally surface at the 0.35 floor and should be triaged as historical learning artifacts rather than live defects unless a second, independent reproducing record is attached. This mirrors the systemic observation recorded for the near-duplicate 3dc2cf closeout and is a triage observation, not a classifier change.