[AnnatarHe]

The Claude review workflow was finishing in a few seconds and posting “No issues found.” This switches .github/workflows/claude-code-review.yml to the built-in high-effort /code-review flow. Only that file changed. The runner stays ubuntu-latest.

This PR’s own review run may be skipped, because the workflow file differs from the default branch. That is expected.

What changed

• Prompt. Replaced the custom review prompt with /code-review high <PR#> --comment (the built-in command, not the code-review@claude-code-plugins plugin).
• Checkout. actions/checkout is now v7 with fetch-depth: 0, so the review can diff, log, and blame against the base branch.
• Permissions. pull-requests and issues are write. Added checks: read, statuses: read, and actions: read, plus additional_permissions: actions: read.
• Tools. claude_args allows Bash(gh:*), Bash(git:*), and mcp__github_inline_comment__create_inline_comment.
• Logs. show_full_output: true, so a run that stops early still leaves a reason in the job log. The comment notes that this repository is public and those logs are visible with the run.
• Concurrency. Group claude-review-<PR#> with cancel-in-progress: true, so a new push replaces a review of a stale diff.
• Triggers. Also runs on ready_for_review and reopened.
• Skip. Drafts and bot PRs are skipped. Drafts are reviewed once marked ready.
• Assertion. New “Assert the review actually ran” step fails a successful action that spent 0 turns and $0, and warns when a review launches no subagents.
• Not set. CLAUDE_CODE_DISABLE_BACKGROUND_TASKS was not in this workflow, so nothing was removed.

  

---

^{Need help on this PR? Tag @codesmith-bot with what you need.}

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.