Reviewer must run tests and detect missing specs before approving
Done
Bug
High
Description
**Quality Control: Reviewer Agent Must Actually Verify Code Works**
**Current Problem:**
Tickets are being "approved" but then failing specs are found after merge. The reviewer agent is not actually verifying the code works before approving.
**Observed Issues:**
- PR #57 submitted with debug dashboard UI but NO feature specs - still got code review PASS
- Failing specs discovered after tickets were marked "approved"
- Reviewer appears to only read code, not run it or test it
**Current Reviewer Behavior (Insufficient):**
1. Reads the code changes
2. Checks style and structure
3. Writes "PASS - No bugs found"
4. Approves ticket
**Required Reviewer Behavior:**
1. **Run the test suite** - `bundle exec rspec` before any approval
2. **Check for missing specs** - UI changes require feature specs
3. **Reject if tests fail** - No approval until all tests pass
4. **Optionally run the app** - For UI changes, actually start rails server and click through
5. **Flag missing test coverage** - Explicitly call out what tests are needed
**Example of What Should Have Happened with PR #57:**
```
Review Summary: REJECT - Missing Tests
Issues:
1. No feature specs provided for debug dashboard UI
- spec/features/debug_dashboard_spec.rb - MISSING
- spec/features/sessions_page_spec.rb - MISSING
- spec/features/terminal_page_spec.rb - MISSING
2. Run test suite: 31 examples, 0 failures
- Only model tests included
- Feature tests required for UI components
Action Required:
Add feature specs before this can be approved. See ticket #95 for rejection details.
```
**Files to Update:**
- Reviewer agent system prompt/instructions
- Review workflow logic
- Code review checklist
**Detection Rules:**
- UI/view changes → require feature specs
- Controller changes → require request/spec
- Model changes → require model specs
- Any new public endpoint → require integration test
Working Memory
{
"assigned_agent_id": 5
}
Ticket Stats
Status:
Done
Priority:
High
Type:
Bug
Comments
0 commentsAdd a Comment
No Subtasks Yet
Break down this ticket into smaller, manageable subtasks