chore(e2e): remove retired terminal proof helpers - #1283
Conversation
Remove the orphaned generator and one-shot fixture after automatic live-proof planning was retired in #1280. Keep historical artifact compatibility and low-level drivers unchanged.
|
🦞👀 Pull request received. I will update this pull request when review starts. |
|
Codex review: needs maintainer review before merge. Reviewed August 28, 2026, 12:26 PM ET / 16:26 UTC. ClawSweeper reviewWhat this changesThis PR removes the unused standalone terminal-proof generator and its one-shot fixture, leaving historical proof records intact. Merge readinessKeep open: this is a focused, viable cleanup that completes the retirement introduced by #1280; no active caller or package command depends on either deleted helper. Priority: P3 Review scores
Verification
How this fits togetherClawSweeper’s automatic live-proof execution has been retired, while the parser, publication path, and historical proof artifacts remain compatible. The deleted scripts were standalone helpers for the retired generation path and no longer feed an active workflow. flowchart LR
A[Review decision] --> B[Retired automatic proof path]
B --> C[Standalone terminal helpers]
D[Historical proof records] --> E[Parser and publication]
C --> F[Deleted unused scripts]
E --> G[Historical compatibility retained]
Before merge
Agent review detailsSecurityNone. Review metrics
Technical reviewBest possible solution: Land the two deletions while retaining the documented historical-artifact parser, verification, publication, and retraction paths. Do we have a high-confidence way to reproduce the issue? Not applicable as a bug reproduction: this is a deletion-only cleanup. The PR body supplies current-head terminal evidence that both retired entry points now fail before invoking Codex or creating output. Is this the best way to solve the issue? Yes. Deleting the two unreferenced helpers is the narrowest maintainable completion of the merged retirement, while the current parser and historical-publication compatibility tests retain the still-supported behavior. AGENTS.md: found and applied where relevant. Codex review notes: model internal, reasoning high; reviewed against 468dab46ff57. LabelsLabel changes:
Label justifications:
EvidenceWhat I checked:
Likely related people:
Rating scale
Overall follows the weaker of proof and patch quality. Workflow
|
Summary
Follow-up to retiring automatic live proof: remove the unused standalone Codex generator and one-shot terminal fixture that were left behind after their planner was deleted.
This is exactly two file deletions, 307 lines removed, zero additions. There is no replacement planner, manual-tool fork, execution toggle, dependency change, or relaxation of an active security boundary.
Ownership and compatibility
Four independent cleanup passes checked reuse, simplification, efficiency, and ownership. Neither helper has a current caller, package command, test dependency, or supported manual-use contract. Historical proof documentation still records the generator's source hash; that provenance is deliberately preserved.
Historical parsing, verification, comment rendering, artifact publication, recording retraction, low-level drivers, production schema/prompt, and workflows are unchanged. OpenClaw Bay is unaffected: no lifecycle, timing, telemetry, presentation, or observer data-contract change. This is internal e2e-tool cleanup, not a user-facing feature change.
Controlled behavior proof
Exercised committed head
b226be81c79f826b3a50b672e13d660691acc63cover basef72ea010c1d7fd134b1bd0826b3a707778c312bc, on macOS with Nodev24.20.0.The removed-entry-point behavior is reproduced with:
Controlled invocations used fresh scratch output paths, HOME/TMPDIR, and a harmless Codex marker executable on PATH. Both correctly exited 1 / MODULE_NOT_FOUND, without invoking Codex or creating the requested output. A tracked-source search found no active references outside immutable historical proof records.
The current production parser and verification/comment renderer also consumed a previously recorded real terminal decision and verification artifact. The historical recommended plan survived parsing unchanged, verification remained valid, and all five PASS markers rendered. New model output remains constrained to the fixed empty
not_applicableshape. No historical command, inference, recording, upload, or publication was executed for this cleanup.Historical input SHA-256:
a7ee1c05499e225e08c301f75d2e9fe80b1309ebfbc6a3af2b38d4f5cc7ef10853715a78327c2d426d52cb746837b33ede92959429e97ce91fc3ff308a15e7bdThese are archived inputs to the current parser/render replay, not a claim of new current-head terminal execution. The retained runtime, schema, prompts, workflows, docs, and dependencies were verified byte-unchanged from base.
Validation
pnpm run checkwas invoked; a complete local pass is not claimed. The baseline has an independently reproduced macOS action-ledger concurrency stall outside these deletions. Its source/tests are untouched. Hosted CI for this PR must pass before merge.Limits: this proves removal of the standalone CLI surfaces and preservation of historical parser/render behavior. It does not claim a new live-proof lane, production publication run, deployment, or model-generation test.