# Direct Jira recovery evidence: ST-2854 and TCID-12973

- Authorized action: deploy the two bounded fixes, skip replay, update the
  existing ticket status metadata and internal comments directly.
- Runtime revision: `e1808bcb0a7f293998b0a9dc25f7e44dab84e88f`.
- Staging effective time: `2026-08-03T10:13:22Z`.
- Runtime proof: Deployment 1/1 updated and Ready; replacement Pod Ready with
  zero restarts; `/healthz` and `/readyz` returned HTTP 200.

## ST-2854

- Jira workflow status before and after: `已解决`.
- Existing internal comment ID: `23354939`.
- Pre-write guard: issue key matched and `public=false`.
- Write: direct Jira platform API `PUT` to the exact comment ID using the
  previously reviewed draft; no replay was run.
- Readback: same comment ID, `public=false`, corrected full-scan/read-amplification
  analysis, and footer `自动回答: 是；人工处理: 否`.

## TCID-12973

- Jira workflow status before and after: `已解决`.
- Existing internal comment ID: `23327373`.
- Pre-write guard: issue key matched and `public=false`.
- Write: direct Jira platform API `PUT` to the exact comment ID using the
  previously verified bounded closure draft; no replay was run.
- Readback: same comment ID, `public=false`, monitoring-causality boundary,
  safe closure acknowledgement, and footer
  `自动回答: 是；人工处理: 否`.

## State boundaries

- No Jira transition, new comment, or customer-visible write occurred.
- Existing DB runs, answer evaluations, feedback, and Base rows were not
  rewritten. The post-rollout DB check found no queued or running jobs and
  confirmed both poll cursor rows remained queryable.
- This evidence records the authorized one-time Jira recovery; it does not
  claim an atomic transaction between Jira and historical analytics views.