claude-code
  • Joined on 2026-10-05
claude-code commented on pull request mi/werkrevisor#139 2026-10-23 16:25:33 +02:00
werkrevisor 0.66.0: draft-review --attach puts the assessment after the report and replaces the report of the same commit

Fixed in 4578286: after a failed edit, attach reads the comment again and deletes only the new SVGs that it does not link; if the read fails too, it keeps them. The old SVGs go only if the new…

claude-code pushed to claude-code/assessment-after-report at mi/werkrevisor 2026-10-23 16:25:33 +02:00
4578286bab --attach keeps the new SVGs when a failed edit may have been applied
claude-code commented on pull request mi/werkrevisor#139 2026-10-23 16:25:33 +02:00
werkrevisor 0.66.0: draft-review --attach puts the assessment after the report and replaces the report of the same commit

Fixed in 4578286: the fake counts asset IDs up and puts the ID into the URL, as Gitea never reuses them; the replacement test now replaces twice and checks the URL of the new asset.

claude-code pushed to claude-code/hypothetical-findings at mi/werkrevisor 2026-10-23 16:19:04 +02:00
a048f26435 Round 5: Scenario#137.01 says that the name stands without quotes
claude-code commented on pull request mi/werkrevisor#137 2026-10-23 16:19:03 +02:00
werkrevisor 0.74.1: hsadmin-ng findings that cannot occur are hints, not ORANGE

Partly fixed in a048f26: the Given of Scenario#137.01 now says that tariffNos stands without quotes. The scenarios stay without SQL terms, since the classifier rates those as implementation…

claude-code commented on pull request mi/werkrevisor#137 2026-10-23 16:19:03 +02:00
werkrevisor 0.74.1: hsadmin-ng findings that cannot occur are hints, not ORANGE

Round 5 (72372b63), answered in a048f26:

  • YELLOW spec, Scenario#137.01 implementation_details 0.62: left to the human. The scenario is about the name of a database field, so it cannot…
claude-code commented on pull request mi/werkrevisor#137 2026-10-23 16:19:03 +02:00
werkrevisor 0.74.1: hsadmin-ng findings that cannot occur are hints, not ORANGE

Does not hold as a gap of this PR: the rules change what the model writes, which no deterministic test can show. The dry runs in the PR document ran with [engine] mode = "combined" of `config/hsa…

claude-code pushed to claude-code/assessment-after-report at mi/werkrevisor 2026-10-23 16:11:12 +02:00
ff0dd748d8 --attach also folds the other reports after a replacement
e653e1767c PR document: help texts of draftreview.py, line length
1e0952bc22 werkrevisor 0.66.0: --attach replaces the report of the same commit and folds earlier ones
b50f3ea5d6 werkrevisor 0.65.3: --checked puts the assessment of the findings after the report
0e74528023 PR document: the assessment of the findings after the report
Compare 6 commits »
claude-code commented on pull request mi/werkrevisor#139 2026-10-23 16:11:12 +02:00
werkrevisor 0.66.0: draft-review --attach puts the assessment after the report and replaces the report of the same commit

Fixed in ff0dd74: a replacement folds the reports of the other heads too; test_a_replaced_report_folds_the_reports_of_other_heads failed before the fix.

claude-code pushed to claude-code/hypothetical-findings at mi/werkrevisor 2026-10-23 15:57:57 +02:00
72372b6304 Round 4: version 0.66.1 after the rebase; scenarios with the example tariffNos
ffb999beb7 The hsadmin-ng rules name the lower-case lookup; dry runs in the PR document
2c481f221d Round 3: quoted references still count; the ruled-out rule overrides the severity rubric
d037b98ea0 Round 2: one case per scenario, in the words of the review
c2ad065d23 Round 1: a ruled-out condition needs proof; scenarios describe the review
Compare 9 commits »
claude-code commented on pull request mi/werkrevisor#137 2026-10-23 15:57:56 +02:00
werkrevisor 0.74.1: hsadmin-ng findings that cannot occur are hints, not ORANGE

Round 4 (1bd10be7), fixed in 72372b6:

  • Merge conflict: rebased onto main with PR#138; version 0.66.1, above PR#139 with 0.66.0.
  • YELLOW spec, Scenario#137.01 and #137.02: both use the…
claude-code commented on pull request mi/werkrevisor#139 2026-10-23 15:32:05 +02:00
werkrevisor 0.66.0: draft-review --attach puts the assessment after the report and replaces the report of the same commit

Fixed in 020d14e: the Solution of the PR document now names the help texts in werkrevisor/draftreview.py.

claude-code commented on pull request mi/werkrevisor#139 2026-10-23 15:32:04 +02:00
werkrevisor 0.66.0: draft-review --attach puts the assessment after the report and replaces the report of the same commit

Fixed in 020d14e: the line is wrapped after a comma.

claude-code pushed to claude-code/assessment-after-report at mi/werkrevisor 2026-10-23 15:31:50 +02:00
020d14eff6 PR document: help texts of draftreview.py, line length
claude-code pushed to claude-code/hypothetical-findings at mi/werkrevisor 2026-10-23 15:08:28 +02:00
1bd10be73d The hsadmin-ng rules name the lower-case lookup; dry runs in the PR document
claude-code pushed to claude-code/codex-status-who at mi/werkrevisor 2026-10-23 15:08:03 +02:00
aedea5d108 A RED head shows author ✍️ instead of queued; the rollout note names WIP PRs
claude-code commented on pull request mi/werkrevisor#138 2026-10-23 15:08:02 +02:00
werkrevisor 0.65.2: the status ai-review/codex names who acts next

Answer to the review of 5caafdf (round 2), fixed in aedea5d:

  • ORANGE "RED Status": fixed, see the line comment.
  • Hint "Rollout Claim": holds; the PR document now says that a WIP PR keeps…
claude-code commented on pull request mi/werkrevisor#138 2026-10-23 15:08:02 +02:00
werkrevisor 0.65.2: the status ai-review/codex names who acts next

Fixed in aedea5d: a RED head now shows author ✍️: werkrevisor found RED; push a commit that fixes the cause; only a missing result shows queued ⏳:. test_a_red_head_gets_no_codex_review…

claude-code commented on pull request mi/werkrevisor#137 2026-10-23 15:05:45 +02:00
werkrevisor 0.74.1: hsadmin-ng findings that cannot occur are hints, not ORANGE

Round 3 (e3417f81), fixed in 308ce5e:

  • ORANGE, blanket suppression: fixed, see its thread.
  • Ambiguous severity: fixed, see its thread.
  • Finding 3, stale diagrams: does not hold, as in…
claude-code commented on pull request mi/werkrevisor#137 2026-10-23 15:05:44 +02:00
werkrevisor 0.74.1: hsadmin-ng findings that cannot occur are hints, not ORANGE

Fixed in 308ce5e: only the case of unquoted identifiers is exempt now. A quoted identifier that the PR adds or refers to, with a case that differs from other code, is still a finding. The…