Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
davidvornholt avatar

Code Review

  • 1 installs
  • Updated May 9, 2026
  • davidvornholt/papersync

Automated code review and quality assessment for pull requests.

About

Code review skill providing automated quality checks and feedback on pull requests. CI-integrated.

  • Automated code review
  • Quality feedback

Code Review by the numbers

  • 1 all-time installs (skills.sh)
  • Ranked #984 of 1,352 Code Review & Quality skills by installs in the Skillselion catalog
  • Data as of Jul 8, 2026 (Skillselion catalog sync)
npx skills add https://github.com/davidvornholt/papersync --skill code-review

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs1
Last updatedMay 9, 2026
Repositorydavidvornholt/papersync

What it does

Automated code review and quality assessment for pull requests.

Files

SKILL.mdMarkdownGitHub ↗

Strict Code Review

Use this skill for local review requests. Review only; do not edit files unless the user explicitly asks for fixes after the review.

Review Posture

  • Act as a strict reviewer looking for problems, regressions, and repo-contract violations.
  • Findings must be grounded in inspected code, diffs, tests, command output, AGENTS.md, matching skills, or documented framework/package behavior.
  • Do not hallucinate requirements, business rules, files, APIs, runtime behavior, tests, command results, or user intent.
  • If evidence is incomplete, either inspect more local context or describe the risk precisely without pretending it is confirmed.

What To Inspect

Before reviewing, gather enough local context to support findings:

  • Read the relevant AGENTS.md instructions.
  • Inspect the description frontmatter for local skills under .agents/skills/*/SKILL.md and follow any that match the reviewed code.
  • Inspect the changed files or requested files, plus nearby callers, tests, package manifests, scripts, and docs as needed.
  • Use git diff, git status, rg, and focused file reads to understand scope.
  • Prefer bun run check from the repo root. If unavailable, inspect package.json and run the closest relevant lint/typecheck/test command.

Finding Categories

Group findings by these sections, ordered by severity within each section:

1. Blocking Findings

  • Any AGENTS.md violation, even small ones.
  • Correctness bugs, runtime crashes, broken builds, security issues, data loss, accessibility failures, invalid types, broken public APIs, missing required env documentation, missing required tests, or architecture boundary violations.
  • Major maintainability issues that should block acceptance, such as misplaced ownership, dependency direction violations, or bad abstractions that make the change unsafe to build on.

2. Non-Blocking Findings

  • Real issues that should be fixed but do not block acceptance.
  • Examples: confusing names, unclear errors, weak but present tests, avoidable duplication, awkward structure, minor performance concerns, local style drift, or maintainability risks.

3. Nits

  • Tiny polish issues and small consistency improvements.
  • Include nits when they exist, but keep them after substantive findings.

Output Contract

  • Use file and line references for every finding whenever possible.
  • Explain the impact and the concrete problem, not just a preference.
  • Keep findings concise and actionable.
  • If no problems are found, say that no review findings were found, summarize what was inspected, and mention checks actually run.
  • If findings exist, end by asking whether you should fix all findings and issues.
  • Mention tests or checks run only if they were actually run. Mention meaningful verification gaps when they affect confidence.

Related skills

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.