// Configure autofix to only fix MAJOR and CRITICAL issues (unattended mode)
| name | reviewer-autofix-ignore-minor-issues |
| description | Configure autofix to only fix MAJOR and CRITICAL issues (unattended mode) |
| allowed-tools | Bash(jq *) |
Run this command:
jq -n --argjson existing "$(cat .reviewer/settings.local.json 2>/dev/null || echo '{}')" '$existing * {"autofix": {"append_to_prompt": "Please autofix as normal, except: 1. Never ask questions. You are running unattended and the user is not there to answer your questions. Instead, think hard about whether to accept each given patch. If you decide *not* to accept it, then create a *new* branch with that fix commit. Call the branch (current_branch_name)___(fix_description) *and be sure to push it remotely* then by sure to check the normal branch back out when you'\''re done. Also be sure to tell the user that you did this. 2. You only *have* to fix MAJOR and CRITICAL issues. If there are issues that you do NOT fix, append the json object(s) for those error(s) that were *not* fixed into .reviewer/outputs/autofix/unfixed/<current-git-hash>.jsonl If so, be sure to mention those issues in your final summary as well."}}' > .reviewer/settings.local.json.tmp && mv .reviewer/settings.local.json.tmp .reviewer/settings.local.json
Then confirm that autofix has been configured to only fix MAJOR and CRITICAL issues.
Review the conversation transcript for behavioral issues (misleading behavior, disobeyed instructions, instructions worth saving).
Assess whether the approach taken on a branch is the right way to solve the problem.
Make the stop hook non-blocking (remind once, then let the agent through)
Restore the stop hook to blocking mode (default: block up to 3 times before letting through)
Set the maximum number of times the stop hook will block before letting the agent through
Automatically find and fix code issues in the current branch. Iteratively verifies, plans fixes, and implements them with separate commits. Defers all review to the end.