Skip to content

JPeetz/agent-skills / skills/code-quality/code-review-and-quality

code-review-and-quality

Verified

Use this skill when reviewing any code change, PR, or diff before merge — whether written by yourself, another agent, or a human. Conducts multi-axis code review across correctness, readability, architecture, security, and performance with structured severity labeling and verification gates.

FileKindSize
CHANGELOG.mdmarkdown529 B
LICENSEother979 B
SKILL.mdskill md11.8 KB
evals/evals.jsonconfig4.1 KB

evals/evals.json

View on GitHub
[
  {
    "id": "crq-001",
    "description": "Pre-merge PR review triggers the five-axis review with severity labeling",
    "input": "Review this PR before merge: it adds a new REST endpoint that writes to the orders table.",
    "shouldTrigger": true,
    "expect": "Review across correctness, readability, architecture, security, and performance; label every finding with a severity (Critical / Nit / Consider / FYI) and end with a verdict.",
    "assertions": [
      "covers all five axes (correctness, readability, architecture, security, performance)",
      "labels findings with severity prefixes",
      "checks input validation and SQL parameterization",
      "ends with an explicit Approve / Request changes verdict",
      "reviews tests before implementation"
    ]
  },
  {
    "id": "crq-002",
    "description": "Oversized change — 1000+ line PR must be flagged and split requested",
    "input": "Please review my 1,200-line PR that refactors auth, adds a new feature, and changes the DB schema.",
    "shouldTrigger": true,
    "expect": "Flag the change as too large to review properly, note it mixes refactor with feature work, and request a split.",
    "assertions": [
      "flags ~1000+ lines as too large",
      "requests the change be split",
      "points out refactoring and feature work are two changes",
      "does not rubber-stamp a review"
    ]
  },
  {
    "id": "crq-003",
    "description": "Secrets or injection in the diff block the merge as Critical",
    "input": "Review this diff: an API key is hardcoded and a SQL query is built by string concatenation with user input.",
    "shouldTrigger": true,
    "expect": "Classify both as Critical severity (security vulnerability), block the merge, and demand fixes.",
    "assertions": [
      "flags the hardcoded secret as Critical",
      "flags string-concatenated SQL as an injection vulnerability",
      "blocks merge until resolved",
      "proposes the remedy (parameterized query, secrets out of code)"
    ]
  },
  {
    "id": "crq-004",
    "description": "Bug-fix PR without a regression test is a red flag the reviewer must push back on",
    "input": "Review this bug fix for the checkout payment bug. The PR only changes the payment service code, no tests.",
    "shouldTrigger": true,
    "expect": "Review both the fix and the absence of a regression test; flag missing regression coverage as a required change.",
    "assertions": [
      "reviews the fix itself",
      "flags the missing regression test",
      "lists missing regression tests as a red flag",
      "requires tests before approving"
    ]
  },
  {
    "id": "crq-005",
    "description": "Dead-code hygiene after a refactor — identify and ask before deleting",
    "input": "After my refactor, is there any dead code I should clean up? I replaced formatLegacyDate with formatDate.",
    "shouldTrigger": true,
    "expect": "Identify the now-unreachable formatLegacyDate and any orphaned references, list them explicitly, and ask before deleting rather than silently removing.",
    "assertions": [
      "identifies unreachable/unused code",
      "lists the dead code explicitly",
      "asks before deleting anything uncertain"
    ]
  },
  {
    "id": "crq-006",
    "description": "Near-miss negative — writing new code from scratch is not a review request",
    "input": "Write a function that validates an email address and returns a normalized version.",
    "shouldTrigger": false,
    "expect": "Do not trigger the review framework — this is feature authoring with no existing change to review.",
    "assertions": [
      "does not run the five-axis review",
      "does not apply severity labeling"
    ]
  },
  {
    "id": "crq-007",
    "description": "Near-miss negative — explaining code is not a pre-merge review",
    "input": "Can you explain what this existing function does step by step? I'm trying to understand the codebase.",
    "shouldTrigger": false,
    "expect": "Do not trigger pre-merge review — an explanation request is not a review of a change.",
    "assertions": [
      "does not produce a review verdict",
      "does not label findings"
    ]
  }
]
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: 471kUpdated 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: 471kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 471kUpdated 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