Audit ledger: not applicable (non-release audit)
Summary
A post-merge audit of shakacode/react_on_rails batch ror-docs-4286-4297-20260701 found that the batch completed correctly, but required closeout evidence was not persisted in a durable/replayable place.
Affected React on Rails PRs:
Product/release risk was low: the merged docs/agent-guidance work looked correct, review threads were resolved, checks passed, and the linked issues were closed. The gap is process replayability.
Evidence
During audit, these were not recoverable from GitHub or private coordination state:
- The named batch entry for
ror-docs-4286-4297-20260701.
- A persisted QA Evidence block with tested heads/range.
- Replayable priority-finding disposition evidence for P2 findings on #4303 and #4306.
Observed commands/results:
PR_BATCH_SKILL_DIR=.agents/skills/pr-batch \
.agents/skills/pr-batch/bin/agent-coord-bounded --timeout 20 \
status --batch-id ror-docs-4286-4297-20260701 --json
Returned batch not found.
script/pr-merge-ledger 4303 --repo shakacode/react_on_rails \
--changelog-classification not_user_visible
script/pr-merge-ledger 4306 --repo shakacode/react_on_rails \
--changelog-classification not_user_visible
Those ledgers reported unknown_priority_finding_disposition unless local temp disposition JSON files were supplied. The fixes existed, but the machine-readable evidence was not persisted somewhere $post-merge-audit could replay without local files or chat memory.
Desired behavior
Update the agent workflow/process so $pr-batch closeout persists enough evidence for $post-merge-audit to replay later, including:
- QA Evidence with tested PR heads or audited range.
- Merge-ledger disposition evidence for P1/P2/Must-Fix findings.
- Batch id / lane state in coordination or a GitHub comment when private coordination cannot persist it.
- A clear durable location for the final QA lane result.
Process Gap Disposition
- Mechanism target: script
- Motivating miss: batch
ror-docs-4286-4297-20260701 merged cleanly, but post-merge audit could not recover required QA/disposition evidence from durable GitHub or coordination state.
- Replay evidence or park reason: replay the commands above against the React on Rails batch; expected future behavior is that the audit can recover complete QA and priority-finding disposition evidence without local
/private/tmp/ror-ledger-disposition-*.json files or chat transcript memory.
- Non-goal: do not add a broad prose-only reminder or require QA comments for batches where QA is explicitly not required.
Audit ledger: not applicable (non-release audit)
Summary
A post-merge audit of
shakacode/react_on_railsbatchror-docs-4286-4297-20260701found that the batch completed correctly, but required closeout evidence was not persisted in a durable/replayable place.Affected React on Rails PRs:
Product/release risk was low: the merged docs/agent-guidance work looked correct, review threads were resolved, checks passed, and the linked issues were closed. The gap is process replayability.
Evidence
During audit, these were not recoverable from GitHub or private coordination state:
ror-docs-4286-4297-20260701.Observed commands/results:
Returned
batch not found.Those ledgers reported
unknown_priority_finding_dispositionunless local temp disposition JSON files were supplied. The fixes existed, but the machine-readable evidence was not persisted somewhere$post-merge-auditcould replay without local files or chat memory.Desired behavior
Update the agent workflow/process so
$pr-batchcloseout persists enough evidence for$post-merge-auditto replay later, including:Process Gap Disposition
ror-docs-4286-4297-20260701merged cleanly, but post-merge audit could not recover required QA/disposition evidence from durable GitHub or coordination state./private/tmp/ror-ledger-disposition-*.jsonfiles or chat transcript memory.