Skip to content

Historical field note

This record preserves prior investigation or planning evidence. It is not a current operating contract; use the numbered book and current runbooks for instructions.

Dream-Finding Assessment: dreamrepair:58afe279d34e186ee4d6d6125532371c

Finding fingerprint: dreamrepair:58afe279d34e186ee4d6d6125532371c Finding kind / scope: failure_pattern, scope project, repo_area mac. Confidence: low, backed by exactly one evidence record (mem_b888960ded04441fba9e84addff484eb). Candidate origin task: task_154f70795b8d48969f4a1b60d3b1f100 — repo_change "Implement src/mac/openclaw_entrypoint.py: atomic per-boot application-ready token", reported FAILED. Parent task: task_2ba873b5d5ce4f61b116a393e340879f (goal: "Investigate low-confidence dream finding: mac"). Assessed by: fleet worker (dream-finding review; no source/test/skill/deploy edits). Baseline commit under audit: "MAC OpenShell sandbox baseline".

Status: NOT ACTIONABLE — stale single-record finding for a module that never landed

The finding is a low-confidence failure_pattern backed by exactly one evidence record and no linked failing test. It derives from a candidate repo_change that was reported FAILED and whose deliverable (src/mac/openclaw_entrypoint.py) was never committed to this repository. There is no current, checked-in deliverable to repair and no reproducible failure in the tree. The correct disposition is closure on an evidence gap, not a source or test change.

Ground Truth: The Candidate Module Does Not Exist

  • src/mac/openclaw_entrypoint.py is absent at HEAD: git cat-file -e HEAD:src/mac/openclaw_entrypoint.py returns "path does not exist in 'HEAD'". It is also absent from the working tree.
  • No test targets it: git ls-files 'tests/*openclaw_entrypoint*' and any equivalent readiness-token test file return nothing. The only entrypoint test present, tests/test_setup_entrypoint.py, exercises the root setup.py installer arg routing — unrelated to a per-boot ready token.
  • No per-boot / application-ready / ready-token mechanism exists under any other name. Repo-wide greps for ready.?token, application_ready, per.?boot, readiness_token, and openclaw_entrypoint across src/mac, tests, and docs return zero matches. The entrypoint hits in src/mac are container/CLI/docker entrypoints, not an OpenClaw boot ready-token.

Ground Truth: Adjacent Convention Is Intact (target shape, if ever built)

The sibling openclaw_* decision-engine modules are present and green, and define the shape any future "per-boot application-ready token" module would follow (a versioned *_SCHEMA string, frozen dataclasses, pure/injectable decision functions, and a matching test module):

  • src/mac/openclaw_checkpoint_gc.py — CHECKPOINT_GC_SCHEMA = "mac.openclaw_checkpoint_gc.v1"; covered by tests/test_openclaw_checkpoint_gc.py (23 tests).
  • src/mac/openclaw_delivery_continuity.py — DELIVERY_CONTINUITY_SCHEMA = "mac.openclaw_delivery_continuity.v1"; covered by tests/test_openclaw_delivery_continuity.py (17 tests).
  • src/mac/openclaw_fleet_rollout.py — ROLLOUT_PLAN_SCHEMA = "mac.openclaw_fleet_rollout.v1"; covered by tests/test_openclaw_fleet_rollout.py (52 tests).

Verification Performed

  • Presence checks: git ls-files and working-tree test -e for the module and its would-be tests; git cat-file -e HEAD:... for HEAD presence.
  • Repo-wide greps (above) for readiness / ready-token / per-boot / entrypoint mechanisms across src/mac, tests, and docs.
  • Full contract gate: scripts/run-contract-tests.sh (with MAC_TEST_COVERAGE=0) — 9217 passed, 4 skipped, exit 0; the integration slice reported 4 passed, exit 0. The adjacent openclaw_* modules are part of that green run. The working tree is clean; this audit adds only this note.
  • Confidence: low. Evidence: a single record with no linked failing test.
  • The referenced deliverable was never landed, so the "failure" is a second-order artifact of a candidate task's FAILED status, not a defect in any checked-in code.
  • Recommended closure reason: not actionable / no current deliverable — close as a stale low-confidence candidate. If a per-boot application-ready token is genuinely wanted, open a fresh implementation task that builds the module to the adjacent openclaw_* convention above, rather than treating this finding as a repair of existing code.