Skip to content

tim-hub/powerball-harness / harness/skills/harness-review

harness-review

Verified

Multi-angle code and plan review with security, scope, and UI profiles. Use when reviewing code, plans, PRs, or running pre-merge quality gates.

references/governance.md

View on GitHub
# Review Governance

## Summary

Only return `APPROVE` when you can say with evidence that there are no critical issues.

## Pass criteria

`APPROVE` conditions:

- Zero critical or major findings
- No contradiction with the spec source-of-truth (`spec_path`) or `spec_skip_reason`
- No contradiction with `Plans.md` task / DoD / Depends
- No evidence of regression in existing behavior, tests, UX, CLI, configuration, docs, or distribution mirrors
- Evidence exists: verification commands, diffs, file:line citations, test results
- No unresolved TeamAgent Debate disagreements

## Severity

| Severity | Meaning | Verdict |
|---|---|---|
| critical | Secret exposure, data destruction, permission destruction, directly causes a release incident | REQUEST_CHANGES |
| major | DoD not met, spec source-of-truth violation, clear regression, dangerous without test execution | REQUEST_CHANGES |
| minor | Quality improves but not a shipping blocker | APPROVE allowed |
| recommendation | Optional improvement | APPROVE allowed |

If findings are minor / recommendation only, blocking is not required.
If blocking, explain concretely why the finding is major, not just minor.

## AskUserQuestion / decision_needed

For decisions that would break things if guessed wrong, use `decision_needed` instead of `REQUEST_CHANGES`.

`decision_needed` examples:

- The spec source-of-truth needs to be changed
- A `Plans.md` DoD / Depends needs to be changed
- The user must choose between security and UX priority
- Whether to preserve or remove backward compatibility is a business decision

Use `AskUserQuestion` when available.
In Codex environments where it is unavailable, emit `decision_needed.v1` to stdout and do not proceed on a guess.

## Side effects

Review default read-only boundary:

- Do not auto-commit even on `APPROVE`
- Do not push just to review
- commit / push / release is the responsibility of `harness-work` / `harness-release` / explicit user request

## Output evidence

Required:

- Target scope
- Review command executed
- Tests executed
- Accepted findings
- Rejected findings
- Clean result or remaining issues
- Pass/fail for spec source-of-truth / Plans.md / regression gate

An `APPROVE` with empty evidence is invalid.
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 194kUpdated Claude CodeCodexVerifiedInstall