ahstn/oceans-llm / actions/review-agent/skills/code-review
code-review
VerifiedReview GitHub pull requests across correctness, maintainability, architecture, security, and performance, with verified findings submitted as GitHub comments.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 8.3 KB |
| agents/openai.yaml | config | 43 B |
| references/axes/architecture.md | markdown | 1.8 KB |
| references/axes/correctness.md | markdown | 2.2 KB |
| references/axes/maintainability.md | markdown | 2.5 KB |
| references/axes/performance.md | markdown | 1.2 KB |
| references/axes/security.md | markdown | 1.6 KB |
| references/behaviour-communication.md | markdown | 2.9 KB |
| references/dead-code-and-simplifying.md | markdown | 1.2 KB |
| references/github-review.md | markdown | 2.5 KB |
| references/synthesis-decisions.md | markdown | 4.4 KB |
references/axes/correctness.md
View on GitHub# Correctness & Robustness Axis
Focus on behavior.
Look for:
- mismatch with the stated task or expected behavior
- broken invariants, state transitions, or lifecycle assumptions
- missing edge-case handling
- missing error-path handling
- races, ordering bugs, stale state, and off-by-one mistakes
- tests that miss the changed behavior or only test the happy path
Do not spend time on style unless it obscures a correctness issue.
## Evidence-led investigation
Establish the reachable trigger, expected contract, changed behavior, and observable impact. Trace callers, guards, schemas, serializers, migrations, cleanup, and downstream consumers. Compare old and new behavior. Try boundary values, absent values, retries, concurrent calls, cancellation, and partial failure. Reject cases excluded by a verified invariant.
Tests must detect the failure they claim to cover. Flag changed tests that accept broken behavior, replace meaningful assertions with truthiness, or duplicate implementation logic so the same bug passes both. Identify the missed behavior; do not ask for coverage based only on a percentage or test count.
For JavaScript and TypeScript, check falsey defaults, unjoined promises, stale closures, shared mutation, runtime schema drift, module exports, and cleanup. A cast, optional chain, or hook warning alone is not a defect. For Python, check mutable defaults, coroutine handling, iterator reuse, transaction/task boundaries, retry side effects, and timezone or precision changes. Framework validation or locally owned mutable state may make these patterns safe.
For GitHub workflows, trace the effective event, checked-out SHA, job conditions, matrices, outputs, artifacts, and scripts loaded by local actions. Look for skipped required work, false success, wrong deployment targets, and producer/consumer mismatches. Intentional optional diagnostics or a failure checked downstream can be safe. Workflow privilege issues belong to the security axis.
For persistence and orchestration, establish what happens after each partial failure and retry. Prefer atomic state transitions where the contract requires them; distinguish database atomicity from external side effects that need idempotency or compensation.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| 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: 471k | Updated | Claude CodeCodex | Verified | Install |
| 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: 471k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |