Qa Regression

Step 2: Build the Regression Checklist

From the issues and handoff notes, compile a complete list of:

• Every acceptance criterion across all tasks in the milestone
• Every behavior that was implemented
• Any must-fix or should-fix items from previous reviews

This is the regression checklist — everything that should still be working.

Step 3: Run Regression Checks

For each item on the checklist:

1. Verify it's still working. Run the relevant tests, check the behavior, or inspect the code.
2. Check for cross-task interference. Did a later task break something an earlier task built? Look for:
   • Shared state modified by multiple tasks
   • API contracts that changed
   • Configuration or environment changes that affect earlier work
   • Test files modified or deleted by later tasks
3. Document the result. Pass (with evidence) or Fail (with details).

If docs/test-plan.md exists, verify coverage against it:

• Which test matrix rows pass?
• Which fail?
• Which were never implemented?

Step 4: Produce the Regression Report

Present findings:

## Regression Report: [Milestone Name]

**Issues in scope:** [issue filenames]
**Total acceptance criteria:** [N]

### Summary
- **Passing:** [N] criteria
- **Failing:** [N] criteria
- **Not testable:** [N] criteria (explain why)

### Passing Criteria
| Issue | Criterion | Evidence |
|-------|-----------|----------|
| [filename] | [criterion] | [test name / manual check] |

### Failing Criteria
| Issue | Criterion | Details | Severity |
|-------|-----------|---------|----------|
| [filename] | [criterion] | [what's broken and why] | [blocker / high / medium] |

### Cross-Task Interference
[Any cases where a later task broke an earlier task's work.]

### Test Coverage Assessment
[If test-plan.md exists:]
- Test matrix rows covered: [N of M]
- Test matrix rows failing: [N]
- Test matrix rows not implemented: [N]
- Coverage gaps: [list]

### New Issues Discovered
[Anything found that wasn't in the original acceptance criteria.]

Step 5: Create Issue Files

For each regression failure, run .cursor/scripts/next-issue-number.sh to get the next available issue number.

Create an issue file in .workflow/issues/backlog/ using the executor prefix convention: .workflow/issues/backlog/swe-bug-[number].md (or the appropriate expert who will fix it).

Template:

# Regression: [Short descriptive title]

**Type:** bug
**Expert:** swe
**Milestone:** [Milestone name]
**Status:** backlog
**Severity:** [blocker / high / medium]

## Description

[What's broken, when it was introduced, and what the expected behavior should be.]

**Regression from:** [original task issue filename]
**Broken by:** [task that likely caused the regression, if identifiable]
**Found by:** `qa-regression` check of [Milestone name]

## Acceptance Criteria

- [ ] [What "fixed" looks like]

## Technical Notes

**File(s):** [affected file paths]

After creating issue files, run .cursor/scripts/update-issues-list.sh to regenerate the issues list.

Step 6: Update Documents

• Add any patterns to .workflow/lessons-log.md (e.g., "tasks that modify shared config should re-run all tests")
• Note in docs/roadmap.md change log that regression was run and how many issues were found
• The regression report feeds into PM's pm-postmortem skill for the milestone