---
title: "Skill reviewer fixture: review, role & definition · Undominated.ai"
canonical: https://undominated.ai/agents/trailofbits-skill-reviewer/
description: "Reviews one Claude Code skill from a loop dispatch, writes nothing, and leaves a rejected ledger finding closed unless the evidence is new."
---

# Skill reviewer fixture: review, role & definition · Undominated.ai

> Reviews one Claude Code skill from a loop dispatch, writes nothing, and leaves a rejected ledger finding closed unless the evidence is new.

[← Explore all agents](/agents/)

TESTING AND REVIEW / Trail of Bits

# Skill reviewer fixture

Reviews one Claude Code skill from a loop dispatch, writes nothing, and leaves a rejected ledger finding closed unless the evidence is new.

 [Use this definition ↓](#setup)[Original source ↗](https://github.com/trailofbits/skills/tree/8e54050ae63c2f62224c35b1616efe71865b4777/plugins/code-improver/tests/fixtures/plugin-dev/agents)

SOURCE REVIEW

 Reviewed 2026-10-14
 Evidence 4 linked sources
 Publisher Trail of Bits
 Licence [CC-BY-SA-4.0 ↗](https://github.com/trailofbits/skills/blob/8e54050ae63c2f62224c35b1616efe71865b4777/LICENSE)
 Revision 8e54050ae63c
 [Read what was—and wasn’t—checked ↓](#review)

“Reviews one skill per dispatch, edits nothing, and reopens a rejected ledger finding only when the evidence is new.”

 [Trail of Bits · upstream description ↗](https://github.com/trailofbits/skills/blob/8e54050ae63c2f62224c35b1616efe71865b4777/plugins/code-improver/tests/fixtures/plugin-dev/agents/skill-reviewer.md) Our analysis follows below.

01 / THE REASONING

## Why this made the selection.

 - The description says it reviews one skill per dispatch and is not for general code review.
- It says not to edit or write files, and a rejected ledger entry stands unless the new evidence falls outside the recorded reason.

### A good fit for

 - An improvement-loop dispatch that hands this profile one skill and that run's findings ledger

### Weigh up before choosing

 - A Markdown profile does not register a native subagent by itself.
- The install command adds the Trail of Bits marketplace. It does not install this agent by itself.
- This review did not execute the resource.
- This path is under the code-improver test fixtures. The plugin README calls tests/ an offline harness and says the skill-improver entry needs the external plugin-dev plugin. This file is not that external agent.
- The nearest fixture README, plugins/code-improver/tests/fixtures/plugin-dev/README.md, returned HTTP 404. The code-improver plugin README is the nearest README that exists.
- Listed tools are Read, Grep, and Glob. The do-not-edit sentence is an instruction. This review did not test whether a host enforces it.
- The marketplace line does not install the external plugin-dev reviewer named in the code-improver README.

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/code-improver/tests/fixtures/plugin-dev/agents/skill-reviewer.md
- plugins/code-improver/README.md
- README.md
- LICENSE

### Our findings

 - The description says it reviews one Claude Code skill per dispatch, checks the run's findings ledger, and is not for general code review.
- It says to report every defect with a severity, including ones that seem minor or unlikely to be fixed.
- It says not to edit or write files. The ledger copy in the prompt is read-only context, and the loop persists the ledger.
- A rejected status stands. Re-filing requires new evidence the recorded reason does not cover, and disagreeing with that reason is not new evidence.
- It says not to soften a severity because the finding was rejected before. Severity states impact, and the ledger records the disagreement.

### Not established by this review

 - The resource was not executed and its tools were not called.
- No skill and no findings ledger were supplied.
- The external plugin-dev skill-reviewer was not fetched.

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/trailofbits/skills/blob/8e54050ae63c2f62224c35b1616efe71865b4777/README.md)

DOCUMENTED COMMAND

 /plugin marketplace add trailofbits/skills Copy command ↗

Copying does not execute this command. It may retrieve a newer version than the reviewed source.

 - The repository README at this pin shows /plugin marketplace add trailofbits/skills as one unindented line. That line adds the marketplace, not this fixture.
- The code-improver README says the skill-improver entry needs the plugin-dev plugin from the claude-plugins-official marketplace. This file is in the local test harness, not that plugin.
- No fetched file contains a single-line command that installs this fixture alone.

### Before you start

 - A host that can read a Markdown agent profile
- A dispatch that supplies one skill and the run's findings ledger, which this file says it expects

### Compatibility

Claude Code

### Read-only review of one dispatched skill

 - Listed tools are Read, Grep, and Glob.
- The body says not to edit or write files and treats the prompt ledger as read-only context.
- It returns findings. It does not grant a fix, commit, or push.

### Cost model

CC-BY-SA-4.0 with attribution and share-alike. No fee is stated for this file, and 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.

 - [skill-reviewer fixture ↗](https://github.com/trailofbits/skills/blob/8e54050ae63c2f62224c35b1616efe71865b4777/plugins/code-improver/tests/fixtures/plugin-dev/agents/skill-reviewer.md) Checked 2026-10-14 https://github.com/trailofbits/skills/blob/8e54050ae63c2f62224c35b1616efe71865b4777/plugins/code-improver/tests/fixtures/plugin-dev/agents/skill-reviewer.md Supports: one-skill dispatch, no edits, and rejected findings stay closed without new evidence
- [Repository root LICENSE ↗](https://github.com/trailofbits/skills/blob/8e54050ae63c2f62224c35b1616efe71865b4777/LICENSE) Checked 2026-10-14 https://github.com/trailofbits/skills/blob/8e54050ae63c2f62224c35b1616efe71865b4777/LICENSE Supports: CC-BY-SA-4.0 Attribution-ShareAlike 4.0 International
- [code-improver README ↗](https://github.com/trailofbits/skills/blob/8e54050ae63c2f62224c35b1616efe71865b4777/plugins/code-improver/README.md) Checked 2026-10-14 https://github.com/trailofbits/skills/blob/8e54050ae63c2f62224c35b1616efe71865b4777/plugins/code-improver/README.md Supports: tests/ is the offline harness, and skill-improver names an external plugin-dev reviewer
- [Repository README ↗](https://github.com/trailofbits/skills/blob/8e54050ae63c2f62224c35b1616efe71865b4777/README.md) Checked 2026-10-14 https://github.com/trailofbits/skills/blob/8e54050ae63c2f62224c35b1616efe71865b4777/README.md Supports: marketplace install line and the same CC-BY-SA-4.0 statement

KEEP COMPARING

## Other approaches to consider.

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

 [### AWS provider maintainer review ↗ A short review persona for AWS provider changes. It says not to open pull requests.](/agents/hashicorp-aws-provider-maintainer/)[### C# Reviewer ↗ Reviews C# changes by running dotnet build and dotnet format verification before reading the diff, then ranks findings from SQL injection and async misuse to nullable reference types and EF Core traps.](/agents/ecc-csharp-reviewer/)[### C++ Reviewer ↗ Reviews C++ changes by running clang-tidy and cppcheck before reading the diff, then ranks findings from raw new/delete, buffer overflows and use-after-free to format-string attacks and integer overflow on untrusted input.](/agents/ecc-cpp-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/)
