---
title: "PR toolkit code reviewer: review, role & definition · Undominated.ai"
canonical: https://undominated.ai/agents/anthropic-pr-code-reviewer/
description: "Pins the pr-review-toolkit code reviewer, which defaults to unstaged git diff and tells itself to drop issues below a confidence cutoff."
---

# PR toolkit code reviewer: review, role & definition · Undominated.ai

> Pins the pr-review-toolkit code reviewer, which defaults to unstaged git diff and tells itself to drop issues below a confidence cutoff.

[← Explore all agents](/agents/)

CODE REVIEW / Anthropic

# PR toolkit code reviewer

Pins the pr-review-toolkit code reviewer, which defaults to unstaged git diff and tells itself to drop issues below a confidence cutoff.

 [Use this definition ↓](#setup)[Original source ↗](https://github.com/anthropics/claude-plugins-official/tree/41f32e407b450055044f167db7ded5e8ff67d203/plugins/pr-review-toolkit/agents)

SOURCE REVIEW

 Reviewed 2026-10-14
 Evidence 3 linked sources
 Publisher Anthropic
 Licence [Apache-2.0 ↗](https://github.com/anthropics/claude-plugins-official/blob/41f32e407b450055044f167db7ded5e8ff67d203/plugins/pr-review-toolkit/LICENSE)
 Revision 41f32e407b45
 [Read what was—and wasn’t—checked ↓](#review)

“Reviews code against CLAUDE.md, defaults to unstaged git diff, and says to report only issues at confidence 80 or above.”

 [Anthropic · upstream description ↗](https://github.com/anthropics/claude-plugins-official/blob/41f32e407b450055044f167db7ded5e8ff67d203/plugins/pr-review-toolkit/agents/code-reviewer.md) Our analysis follows below.

01 / THE REASONING

## Why this made the selection.

 - The catalogue has feature-dev's code-reviewer at a different path and not this pr-review-toolkit agent.
- The file sets its own reporting cutoff and pins model to opus, which is a host behavior a catalogue can state without treating the cutoff as a measured quality score.

### A good fit for

 - A Claude Code review of unstaged changes when the project rules live in CLAUDE.md
- Supply the intended diff scope and applicable repository instructions.
- Receive high-confidence issue candidates with locations, concrete impact and suggested fixes.

### Weigh up before choosing

 - This review did not execute the resource.
- The confidence bands are instructions in the file. They are not a score this review assigned to the agent.
- The root README install line uses a {plugin-name} placeholder. It does not name pr-review-toolkit by itself.
- This commit is 41f32e407b450055044f167db7ded5e8ff67d203, later than the pin used for other agents already proposed from this repository.
- The confidence score is model self-assessment, not calibrated defect probability, and an empty report is not proof that tests or runtime behavior passed.

02 / THE REVIEW RECORD

## What we actually inspected.

Source review has boundaries. A clear record is more useful than a “safe” badge.

### Material inspected

 - plugins/pr-review-toolkit/agents/code-reviewer.md
- plugins/pr-review-toolkit/LICENSE
- README.md
- Round 2: pinned definition, output contract and permission/dependency boundaries reread on 2026-10-14

### Our findings

 - The agent says to review unstaged changes from git diff unless the user names a different scope.
- It sets model to opus and tells itself to report only issues with confidence at or above 80, grouped as critical from 90 to 100 and important from 80 to 89.
- It says to check project rules, bugs, and significant quality issues, and to confirm briefly when nothing clears that cutoff.
- The default scope is unstaged changes unless the user supplies a different target.
- The reporting threshold suppresses lower-confidence findings and asks for guideline rationale or a direct bug explanation.

### Not established by this review

 - The resource was not executed and its tools were not called.
- git diff was not run for this review.
- The agent was not spawned.

The review applies to the material and revision named here. A newer upstream release can change its behavior.

03 / PUT IT TO WORK

## Use the role in your project.

[Upstream setup instructions ↗](https://github.com/anthropics/claude-plugins-official/blob/41f32e407b450055044f167db7ded5e8ff67d203/README.md)
 - The README says to run /plugin install {plugin-name}@claude-plugins-official or to browse /plugin > Discover.
- The placeholder is not a complete install line for this agent, so install.command is null.
- The agent file is plugins/pr-review-toolkit/agents/code-reviewer.md. The plugin LICENSE at this commit is Apache-2.0.
- Make the diff scope explicit and confirm inherited tool permissions; independently reproduce material findings before accepting them.

### Before you start

 - Claude Code
- A git checkout if the default unstaged-diff scope is used

### Compatibility

Claude Code

### Host-spawned reviewer with a default git diff scope

 - Read the agent prompt
- Inspect the diff the host passes in
- Suggest fixes in its report; the file does not grant a merge or push

### Cost model

Apache-2.0; the agent was not run

04 / FOLLOW THE EVIDENCE

## The source trail.

Our notes are separate from the original resource. Check upstream before adopting a new version.

 - [pr-review-toolkit code-reviewer ↗](https://github.com/anthropics/claude-plugins-official/blob/41f32e407b450055044f167db7ded5e8ff67d203/plugins/pr-review-toolkit/agents/code-reviewer.md) Checked 2026-10-14 https://github.com/anthropics/claude-plugins-official/blob/41f32e407b450055044f167db7ded5e8ff67d203/plugins/pr-review-toolkit/agents/code-reviewer.md Supports: diff scope, model pin, and reporting cutoff
- [pr-review-toolkit LICENSE ↗](https://github.com/anthropics/claude-plugins-official/blob/41f32e407b450055044f167db7ded5e8ff67d203/plugins/pr-review-toolkit/LICENSE) Checked 2026-10-14 https://github.com/anthropics/claude-plugins-official/blob/41f32e407b450055044f167db7ded5e8ff67d203/plugins/pr-review-toolkit/LICENSE Supports: Apache License Version 2.0
- [Repository README ↗](https://github.com/anthropics/claude-plugins-official/blob/41f32e407b450055044f167db7ded5e8ff67d203/README.md) Checked 2026-10-14 https://github.com/anthropics/claude-plugins-official/blob/41f32e407b450055044f167db7ded5e8ff67d203/README.md Supports: placeholder install line

KEEP COMPARING

## Other approaches to consider.

Related by category or shared topics. These are alternatives to inspect, not a measured quality order.

 [### Silent failure hunter ↗ Looks for swallowed errors and bad fallbacks, and requires an output format.](/agents/ecc-silent-failure-hunter/)[### Code slice worker ↗ Pins the Trailmark worker that answers from one source packet, returns a single JSON object, and says it cannot read the repository.](/agents/trailofbits-code-slice-worker/)[### Expert Rust Code Reviewer ↗ Reviews a Rust diff for ownership, unsafe invariants, concurrency, error handling and public APIs, with compiler or Clippy checks for testable findings.](/agents/buffa-rust-code-reviewer/)

 [AI Tools ↗](/tools/)[Skills ↗](/skills/)[Agents ↗](/agents/)[MCP Servers ↗](/mcp-servers/)[Workflows ↗](/workflows/)

## Continue your investigation

 - [Find reusable instructions](/skills/)
- [Inspect connections](/mcp-servers/)
- [Build a toolkit](/tools/)
- [Choose the model](/compare/)
