New: 97,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

On-Demand Code Review

by VeryGoodOpenSource in VeryGoodOpenSource/vgv-wingspan

Runs parallel review agents over a branch, chosen paths or a whole project and merges their findings into one numbered report you can act on by id.

MITAuto-check passedDevelopment

Install On-Demand Code Review

skills CLI
$ npx skills add VeryGoodOpenSource/vgv-wingspan --skill review -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install VeryGoodOpenSource/vgv-wingspan review --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/VeryGoodOpenSource/vgv-wingspan.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/review .claude/skills/review && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
review
GitHub stars
108
Token cost
~1.7k tokens
SKILL.md length
856 words
Files
6 (incl. scripts, references)
Skills in repo
11
Repo updated
First seen
Licence
MIT

At a glance

Runs parallel review agents over a branch, chosen paths or a whole project and merges their findings into one numbered report you can act on by id.

  • Works in 4 steps: Detect Scope → Run Reviews → Consolidate & Present → …
  • Checking a branch before merging it
  • SKILL.md covers Review Scope, Step 1 — Detect Scope, Step 2 — Run Reviews and Step 3 — Consolidate & Present, plus 3 more sections
  • Runs Shell scripts from its folder

What it does

Scope comes from file paths you pass, from a detection script that lists the files changed on the current branch, or, on the default branch, from a question asking whether to review chosen paths or the entire project. Each run gets its own folder under docs/code-review named after a scope slug, so one review never overwrites another branch's report.

Default review agents run in parallel against the Very Good Ventures standards for architecture, tests and simplicity, and a project can add its own agents in CLAUDE.md or replace the defaults. Raw per-agent reports are consolidated into a single report with stable numbered findings, and a reference covers filing findings on a pull request. It needs an agent-capable host, with gh or glab for pull request work.

When your agent uses it

  • Checking a branch before merging it
  • Assessing an existing codebase against quality standards
  • Reviewing hand-written code in one specific directory
  • Turning review findings into numbered items you can pick from

Example prompts

  • “Review the changes on this branch before I open a pull request.”
  • “Run a code review on lib/auth and give me one numbered report.”
  • “Review the whole project, then file the findings on the PR.”

Requirements

  • An agent host that supports parallel subagents
  • gh or glab for the pull request steps
  • Compatibility (from SKILL.md): Designed for Claude Code (or similar products with agent support)
  • Pre-approved tools (allowed-tools): Bash(*/scripts/detect-review-scope.sh), Bash(gh *), Bash(glab *)

Workflow steps

4 steps, taken from the step headings in SKILL.md.

  1. Detect Scope
  2. Run Reviews
  3. Consolidate & Present
  4. Act

What it can do on your machine

Read from SKILL.md and the folder at commit 19e0695. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves these tools, so the agent can use them without asking each time:

    • Bash(*/scripts/detect-review-scope.sh)
    • Bash(gh *)
    • Bash(glab *)

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/ (Shell), which the agent can run.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

  • Compatibility

    Designed for Claude Code (or similar products with agent support)

    From compatibility in the SKILL.md frontmatter.

Context cost

On-Demand Code Review loads about 1.7k tokens when it runs, and up to ~1.8k if it reads all its reference files. Until then it costs about 42 tokens; SKILL.md has 856 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~42
When it runs · the whole SKILL.md, loaded when a task matches
~1.7k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~1.8k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from VeryGoodOpenSource/vgv-wingspan at commit 19e0695, republished under its MIT licence (© VeryGoodOpenSource). 856 words, ~1,727 tokens.

Download SKILL.mdSave it as .claude/skills/review/SKILL.md (or your agent's skills folder). This skill also uses 5 other files; get the full folder from GitHub.
name
review
description
Runs quality review agents on demand — reviews code against VGV standards for architecture, tests, and simplicity, then writes one consolidated, numbered report.
allowed-tools
Bash(*/scripts/detect-review-scope.sh), Bash(gh *), Bash(glab *)
compatibility
Designed for Claude Code (or similar products with agent support)
user-invocable
true
when_to_use
Use when user says "review this code", "review my code", "code review", "review", "check this code", or "review before merging".
argument-hint
[path/to/files/or/directories (optional)]
effort
high

Review code on demand

Run quality review agents. Review manually written code, assess existing codebases, or check a branch before merging. Output is one consolidated report with stable, numbered findings the user can act on by id.

Review Scope

<review_scope>$ARGUMENTS</review_scope>

Step 1 — Detect Scope

Parse the review scope above for optional file paths or directories.

If paths are provided:

  1. Validate each path exists (split on whitespace, check each token).
  2. Use provided paths as review scope. Derive the scope slug deterministically from the first path: drop any file extension, then replace / with - (lib/auth/ → lib-auth, lib/auth/token.dart → lib-auth-token). Same paths always give the same slug.
  3. Announce scope to the user and proceed to Step 2.

If no paths provided:

Run the scope detection script:

bash
${CLAUDE_SKILL_DIR}/scripts/detect-review-scope.sh
  • If SCOPE=branch: use the listed files as scope. The scope slug is the current branch name with / replaced by -. Announce scope summary (changed-file count, areas affected) and proceed to Step 2.
  • If SCOPE=default: tell the user "You're on <CURRENT_BRANCH>. No branch diff available." Use AskUserQuestion: "What would you like to review?" with options:
    • Specify files or directories: accept paths; derive the slug from the first path as above.
    • Review entire project: no scope constraint; slug project.

Step 2 — Run Reviews

Run pwd and let <PWD> be the result — subagents may change directories, making relative paths unreliable. Each run gets its own directory <PWD>/docs/code-review/<slug>/, so raw per-agent reports go in <PWD>/docs/code-review/<slug>/raw/ (absolute) and one run never clobbers another branch's kept report.

Run the default review agents below in parallel. Projects may add agents in their CLAUDE.md (include them alongside the defaults) or replace the default set entirely.

Each agent prompt must include:

  1. The scope constraint — changed-file list, specific paths, or no constraint.
  2. The review agent instructions with <RAW_DIR> set to <PWD>/docs/code-review/<slug>/raw and <name> set to the agent's report name below (a bare stem — the agent writes <RAW_DIR>/<name>.md). Substitute <PWD> and <slug> with their resolved values — do not pass a relative path.

Default agents and their report names (<name>):

AgentReport name
@vgv-review-agentvgv-review
@architecture-review-agentarchitecture-review
@test-quality-review-agenttest-quality-review
@code-simplicity-review-agentcode-simplicity-review

If an agent fails: note it, continue with the successful agents, and record the failure in the report header and chat summary so the user knows the review is incomplete. Offer to retry.

Step 3 — Consolidate & Present

Follow the review consolidation procedure:

  1. Collect every agent's structured findings, deduplicate, order deterministically, and assign stable FINDING-NN ids.
  2. Write one consolidated file to <PWD>/docs/code-review/<slug>/review.md using the report template.
  3. Print the aligned chat summary: lead with the report path and severity counts, then reprint the Critical and Important rows verbatim (same ids, order, titles) and collapse Suggestions to a count.

If no findings: write the short all-clear report and tell the user the code looks good.

Show full SKILL.md (397 more words)Show less

Step 4 — Act

This skill is advisory — take no action until the user picks one. Use AskUserQuestion to present post-review options (the ids and rules follow the consolidation procedure's "Acting on findings" section):

  • Fix critical issues: address every Critical finding by id, then run the project's linter and test runner. One attempt per fix; if validation fails, report what failed and move on. Only modify files within the original review scope.
  • Fix critical + important: same, plus Important findings.
  • Fix specific findings: accept ids from the user (e.g. "FINDING-01, FINDING-04"), or a rule id to act on a whole class (e.g. "fix every tests/missing-test-file").
  • File findings on the PR as comments: follow file findings on the PR — a two-step flow that first asks which findings to include, then how to deliver them (inline comments, one summary comment, or print the drafts for the user to post manually).
  • Keep report and exit: the report stays at docs/code-review/<slug>/ for manual review.

After fixing (if chosen): re-run linter + test runner (no agent re-run), then present a brief summary of which findings (by id) were fixed.

Gotchas

  • Each run writes to its own docs/code-review/<slug>/ directory (report + raw/). Re-running the same scope overwrites only that slug's directory; a different branch or path uses a different slug, so a report you keep is never clobbered by a later run of another scope.
  • Because ids come from a deterministic sort (severity → file → line → rule), re-running on unchanged code produces the same ids — FINDING-03 keeps pointing at the same issue. Each finding also carries a stable rule id (e.g. vgv/missing-null-check) the user can act on as a class.
  • On the default branch with no diff, scope is ambiguous. The skill asks; do not default to reviewing the whole project without confirmation.
  • Agent failures are non-fatal. Always report which agents failed so the user knows the review is incomplete.
  • Auto-fix only touches files within the original scope. If a fix needs changes outside scope, flag it instead of silently expanding scope.

Important

  • One consolidated report per run. Per-agent raw reports live in docs/code-review/<slug>/raw/ for drill-down and are linked from the consolidated file.
  • Reports are untracked working files. Commit or delete them when no longer needed.
  • This skill is advisory. It presents findings and lets the user decide what to act on.
  • When in doubt about a finding, read its linked raw report for full detail before deciding.

© VeryGoodOpenSource, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 5 other files (scripts, references) in skills/review of VeryGoodOpenSource/vgv-wingspan.

  • SKILL.md
  • references/file-findings-on-pr.md
  • references/review-agent-instructions.md
  • references/review-consolidation.md
  • references/review-report-template.md
  • scripts/detect-review-scope.sh

Open the folder on GitHubat commit 19e0695

Compare with similar skills

On-Demand Code Review next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

On-Demand Code Review compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
On-Demand Code Review this skillVeryGoodOpenSource/vgv-wingspan108—~1.7kAutomated safety check: PassMIT
WooCommerce Code Reviewwoocommerce/woocommerce11k3 repos~1.1kAutomated safety check: PassCustom licence
Skill Doli Code ReviewDolibarr/dolibarr7.7k1 repos~1.1kAutomated safety check: PassMIT
Dignified Python Standardsdocling-project/docling68k—~1.5kAutomated safety check: PassApache-2.0
Clean Code GuardamElnagdy/guard-skills1.3k2 repos~4.3kAutomated safety check: PassMIT
Archify Reviewtt-a1i/archify79k—~415Automated safety check: PassMIT

Similar skills

  • WooCommerce Code Review

    woocommerce/woocommerce

    Reviews WooCommerce code changes against the project's standards, flagging backend PHP architecture, naming, documentation, data integrity and testing violations.

    11k GitHub starsUsed in 3 repos~1.1k tokens
    DevelopmentAuto-check passed

  • Skill Doli Code Review

    Dolibarr/dolibarr

    Reviews Dolibarr PHP code for compliance with coding standards and security best practices, and fixes identified issues.

    7.7k GitHub starsUsed in 1 repo~1.1k tokens
    DevelopmentAuto-check passed

  • Dignified Python Standards

    docling-project/docling

    Applies opinionated production Python conventions chosen by the project's Python version: modern type syntax, pathlib, explicit checks and interface guidance.

    68k GitHub stars~1.5k tokensUpdated yesterday
    DevelopmentAuto-check passed

  • Clean Code Guard

    amElnagdy/guard-skills

    Reviews generated or changed production code against Clean Code, SOLID, DRY, KISS, YAGNI and LLM-specific failure modes before it ships, in any language.

    1.3k GitHub starsUsed in 2 repos~4.3k tokens
    DevelopmentAuto-check passed

  • Archify Review

    tt-a1i/archify

    Review Archify issues, PRs, or code through value, cost, and impact to support evidence-based maintenance decisions. Use for issue triage, change reviews, and…

    79k GitHub stars~415 tokensUpdated today
    DevelopmentAuto-check passed

  • Code Review Skill

    awesome-skills/code-review-skill

    Provides comprehensive code review guidance for React 19, Vue 3, Angular 17+, Svelte 5, Rust, TypeScript, Java, Java 8, PHP, Ruby, Rails, Python, Django, FastAPI, Go, C/.NET, Kotlin, Swift, Dart…

    2.1k GitHub stars~2.8k tokensUpdated 29 days ago
    DevelopmentAuto-check: notes

More from VeryGoodOpenSource/vgv-wingspan

All 11 skills in this repo

  • Git Rebase onto Base Branch

    VeryGoodOpenSource/vgv-wingspan

    Rebases the current feature branch onto the latest main, master or develop, stashing uncommitted work first and aborting cleanly if conflicts appear.

    108 GitHub stars~786 tokensUpdated 7 days ago
    Auto-check passed

  • Feature Brainstorm

    VeryGoodOpenSource/vgv-wingspan

    Clarifies what to build before how, by asking one question at a time about a feature idea and then handing the result on to planning.

    108 GitHub stars~1.8k tokensUpdated 7 days ago
    Auto-check passed

  • Create Pull Request

    VeryGoodOpenSource/vgv-wingspan

    Stages, commits, pushes and opens a pull request with a Conventional Commits message, after running the project's checks unless told to skip them.

    108 GitHub stars~1.3k tokensUpdated 7 days ago
    Auto-check: notes

  • Post-Incident Debrief

    VeryGoodOpenSource/vgv-wingspan

    Produces a blameless post-incident debrief with timeline, root cause and follow-up actions after an outage, failed release or significant bug, while details are fresh.

    108 GitHub stars~1.9k tokensUpdated 7 days ago
    Auto-check passed

  • Targeted Emergency Bug Fix

    VeryGoodOpenSource/vgv-wingspan

    Applies a minimal fix to an emergency bug through triage, root-cause location, a hotfix branch and a blast-radius check, with tests and review still required.

    108 GitHub stars~1.9k tokensUpdated 7 days ago
    Auto-check passed

  • Plan

    VeryGoodOpenSource/vgv-wingspan

    Turns high-level brainstorming and ideas into well-structured, actionable implementation plans.

    108 GitHub stars~2.8k tokensUpdated 7 days ago
    Auto-check passed

Categories

Questions about On-Demand Code Review

What does On-Demand Code Review do?

Runs parallel review agents over a branch, chosen paths or a whole project and merges their findings into one numbered report you can act on by id. Scope comes from file paths you pass, from a detection script that lists the files changed on the current branch, or, on the default branch, from a question asking whether to review chosen paths or the entire project. Each run gets its own folder under docs/code-review named after a scope slug, so one review never overwrites another branch's report.

When should I use On-Demand Code Review?

On-Demand Code Review fits situations like: checking a branch before merging it; assessing an existing codebase against quality standards; reviewing hand-written code in one specific directory; turning review findings into numbered items you can pick from.

How do I install On-Demand Code Review in Claude Code?

Run `npx skills add VeryGoodOpenSource/vgv-wingspan --skill review -a claude-code`. Or copy the skill folder (skills/review in VeryGoodOpenSource/vgv-wingspan) into .claude/skills/review in your project. Claude Code loads it when a task matches its description.

How do I install On-Demand Code Review in Codex?

Run `npx skills add VeryGoodOpenSource/vgv-wingspan --skill review -a codex`. Or copy the skill folder (skills/review in VeryGoodOpenSource/vgv-wingspan) into .agents/skills/review in your project. Codex loads it when a task matches its description.

Can I use On-Demand Code Review in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add VeryGoodOpenSource/vgv-wingspan --skill review -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/review, .gemini/skills/review, .github/skills/review and .opencode/skills/review in your project.

What does On-Demand Code Review need to run?

Going by SKILL.md and its folder, On-Demand Code Review needs a shell for the scripts in its folder. Our summary lists: An agent host that supports parallel subagents; gh or glab for the pull request steps. Its frontmatter pre-approves these tools: Bash(*/scripts/detect-review-scope.sh), Bash(gh *), Bash(glab *). Compatibility (from SKILL.md): Designed for Claude Code (or similar products with agent support).

Does On-Demand Code Review access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is On-Demand Code Review safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does On-Demand Code Review use?

On-Demand Code Review is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does On-Demand Code Review use?

About 1.7k tokens (SKILL.md is roughly 6.9k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 49 tokens, read only when the agent opens those files.

What are the alternatives to On-Demand Code Review?

Skills that share tags, products or a category with On-Demand Code Review: WooCommerce Code Review (woocommerce/woocommerce, 11k stars), Skill Doli Code Review (Dolibarr/dolibarr, 7.7k stars), Dignified Python Standards (docling-project/docling, 68k stars) and Clean Code Guard (amElnagdy/guard-skills, 1.3k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains On-Demand Code Review?

VeryGoodOpenSource (a GitHub organization) maintains it in VeryGoodOpenSource/vgv-wingspan, which has 108 GitHub stars. The repository holds 11 skills in this directory. The repository was last updated on September 30, 2026.

Source: VeryGoodOpenSource/vgv-wingspan on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.