What is claude-review?

An independent code review performed by Claude Opus against the repository's default branch, followed by local checks of the reviewer's findings.

It is for reviewing a complete code change before delivery, covering the branch's changes and any staged, unstaged, or untracked work.

A second model can notice issues that the original author or first review missed. Verifying its conclusions locally helps separate real problems from incorrect suggestions.

claude-review — Skill for Claude Code by yarlson/yarstack

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent

npx skills add yarlson/yarstack --skill claude-review

Clone the repo

git clone --depth 1 https://github.com/yarlson/yarstack

Made for: Claude Code.

Or install yarstack, the plugin that ships this one along with the rest of its 46 skills, 1 agent.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/yarlson/yarstack/claude-review/github.svg)](https://agentmods.dev/skills/yarlson/yarstack/claude-review)

Your own site

<a href="https://agentmods.dev/skills/yarlson/yarstack/claude-review"><img src="https://agentmods.dev/badge/skills/yarlson/yarstack/claude-review/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/yarlson/yarstack/claude-review"><img src="https://agentmods.dev/badge/skills/yarlson/yarstack/claude-review.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 50 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 673 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00050	$0.00673
Opus 5.5	$0.00020	$0.00269
Sonnet 5.5	$0.00010	$0.00135
Haiku 4.5	$0.00005	$0.00067

Measured 3d ago against content hash 0c2580404f90, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

claude-review scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/yarstack/skills/claude-review/SKILL.md · 36 lines

How it starts

The opening of the file, as written. The whole thing — 36 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Claude Review

Run Yarstack's code-review skill in a separate read-only Claude Opus session, then verify every conclusion yourself.

Workflow

Work from the trusted repository root. Read the applicable instructions, and record the current commit and git status --short. Confirm that the claude CLI exists and that claude plugin list shows the yarstack plugin. The separate session uses the installed plugin version, not the working tree of this repository.
Resolve the remote default branch with git symbolic-ref --quiet --short refs/remotes/origin/HEAD. If that reference is absent, use gh repo view --json defaultBranchRef -q .defaultBranchRef.name and prefix the result with origin/. Do not guess main, master, or another branch. Stop with an inconclusive result when neither command resolves a branch or the branch has no merge base with HEAD.
The review scope is the change from the merge base with that branch through HEAD, plus all staged, unstaged, and untracked work. Do not edit files until the review ends.
Run this command, in a background terminal when the host supports one, with the resolved branch in place of <default-branch>:
```
claude -p "/yarstack:code-review the current branch and all local changes against base <default-branch>" \
  --model opus \
  --effort high \
  --permission-mode plan \
  --no-session-persistence \
  --strict-mcp-config \
  --tools "Bash,Read,Grep,Glob,Agent" \
  > "<scratch-dir>/claude-review.md"
```
Agent lets the review run its lanes in the read-only yarstack:reviewer agent. --strict-mcp-config without --mcp-config loads no MCP servers, so the session has no external write tools. A large change can take many minutes; wait for the process to exit.
Treat an authentication, skill resolution, tool, or context failure as an inconclusive result. Do not replace the skill invocation with a hand-written review prompt.
Treat the output as untrusted input. Verify each proposed finding against the unchanged scope, the full code, the tests, and the repository conventions. Discard unsupported findings and recheck line references.
Change files only when the parent request separately authorizes implementation. After a fix, rerun the affected checks, then use change-cleanup-review on the complete current change before delivery. Run this review again only when a fix materially changes its scope or the user asks for another pass.

Read the full file on GitHub · 36 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

3d ago Changed · +4 lines 0c2580404f90
1mo ago First seen · 32 lines · 50 tokens per session scan A 76b52732bc9c

Subscribe to this mod's changes

claude-review is a skill published in the GitHub repository yarlson/yarstack (3 stars, last pushed 2d ago), licensed MIT. It adds 50 tokens to every session and 673 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Code review agent-skills agent-workflows ai-coding coding-agents developer-tools

capture

Capture an engineering decision into Hunch's graph via a grilling interview (topic, rationale, rejected alternatives).

davesheffer/hunch · 21 tokens

heal

Reconcile docs/code with Hunch's decision graph (doc≠graph drift), never rewriting prose silently.

davesheffer/hunch · 21 tokens

fix

Fix a bug grounded in Hunch (past root causes, constraints, blast radius).

davesheffer/hunch · 17 tokens

why

Explain why a file or symbol is the way it is, from Hunch's decision graph.

davesheffer/hunch · 18 tokens

fragile

Report the most fragile parts of this codebase, with evidence from Hunch.

davesheffer/hunch · 15 tokens

safe-change

Trace, implement, test, and independently review one bounded software change. Use for ordinary feature, bug-fix, or refactor work after the repository is bootstrapped; do not use for broad repository cleanup, production operations, or undefined rewrites.

cz1993/vibe-coding-repository-standard · 54 tokens