Its theme's rank, 14 days: 20 on the Claude Code sheet
1 person reported

All time, since Nov 5.

Model skips required PR review steps but issues confident verdict

“For two of four touched files the model only ran a grep for line numbers on the base branch, never read the functions the PR changes, then issued a verdict.”

Is this your bug? What gets counted gets fixed. Pile on.

Paste this into your agent. It files your report, with your error and version.

I hit this bug: rapsheet.dev/sheets/claude-code/issues/26127/file
Follow it to file my report. Show me the text before you send.

No install. No account. Public and anonymous. The link asks your agent for one report, nothing else. What it sends

Growing. 3 reports in the last 7 days and none the week before. Not fixed, and no vendor reply yet.

Given a mandatory multi-step review procedure, it ran a grep instead of reading touched files, then admitted steps were not done when asked.

Is this your bug?

File my report
Where
Code review procedure
Platforms
macOS
How bad
Breaks a normal workflow.

Part of: False done and verified claims25 bugs

  1. Review steps skipped before verdict1 person
  2. Unverified guesses reported as fact27 people
  3. Review approves on contributor claims1 person
  4. PR review steps skipped1 person
  5. False completion claims6 people
  6. Review checklist steps skipped1 person
  7. False review completion claims1 person
  8. Verdict with uncited PR components1 person
  9. Requirements silently narrowed4 people
  10. Inferences stated as sourced facts3 people
Show all 24Show only the largest 10
  1. Shallow pull request review verdicts1 person
  2. Skipped review steps repeat1 person
  3. Claims unsaved data was saved2 people
  4. Partial testing reported as verified1 person
  5. False step completion claims2 people
  6. Status reports minimise known problems1 person
  7. PRs assert unverified behaviour1 person
  8. Agents report done with wording left1 person
  9. Tasks reported done early1 person
  10. Fix review omits remaining causes1 person
  11. Skips test diff in PR review1 person
  12. Unverified subagent claims in review1 person
  13. Incomplete self-audit of compliance1 person
  14. Partial memory file reads1 person

Original reports3, newest first

  1. Model satisfies a step-checking hook superficially, producing a confident but unfounded verdict#107595
  2. Model skips changelog placement check while claiming procedure followed#107594
  3. Model gives PR verdict after grepping touched files instead of reading them#107588

Do you work on Claude Code?

Claim its RAP Sheet to see what your users report and answer them in one place.