skills.rest
lukasdeveloperacc
lukasdeveloperaccCOMMUNITY
Explore all skills by this author →
GitHub

harness-reviewer

Review code diffs and emit structured STOP/GO verdicts for the Claude cmux harness loop.

Updated Jun 7, 2026
Search Tags:#automation#code review#Codex#structured verdict#AI reviewer
One-click install
npx skills add https://github.com/lukasdeveloperacc/PersonalAgentPlugins --skill harness-reviewer-lukasdeveloperacc
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: harness-reviewer
Source: https://github.com/lukasdeveloperacc/PersonalAgentPlugins/tree/main/codex/plugins/socrates-codex-partner-plugin/skills/harness-reviewer
Command: npx skills add https://github.com/lukasdeveloperacc/PersonalAgentPlugins --skill harness-reviewer-lukasdeveloperacc

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill automates the review process for the Claude cmux harness loop, providing structured STOP/GO verdicts without editing source files.

Core Features & Use Cases

  • Automated Reviewing: Reviews diffs, requirements, tests, and architecture, emitting a structured verdict.
  • File-Authoritative: Writes review outputs to designated files, ensuring file integrity.
  • Use Case: Ideal for reviewing code changes in the Claude cmux harness loop, providing immediate feedback on code modifications.

Quick Start

Trigger the harness-reviewer skill and provide the necessary inputs such as loop_id, attempt, and target verdict path.

Frequently Asked Questions about harness-reviewer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate code review to generate structured STOP or GO verdicts?▼

To automate code review for structured STOP or GO verdicts, this Skill reviews code diffs, requirements, tests, and architecture, then emits a structured verdict without modifying source files.

What inputs do I need to trigger an automated AI code review in a harness loop?▼

Triggering an automated AI code review in a harness loop requires providing the loop_id, attempt number, and target verdict path to generate the corresponding review markdown and done marker.

How does the code review verdict mechanism work for automated loops?▼

The code review verdict mechanism works by analyzing diffs and writing file-authoritative outputs, generating a REVIEW markdown file, a .done marker, and printing a HARNESS_VERDICT_DONE sentinel.

Can I use this automated reviewer to evaluate code changes without modifying my source files?▼

Yes, you can use this automated reviewer to evaluate code changes without modifying source files, as it is designed to provide immediate feedback by writing review outputs exclusively to designated files.

What are the limitations of using a file-authoritative structured verdict for code review?▼

A limitation of using a file-authoritative structured verdict for code review is that it depends on predefined inputs like loop_id and attempt, restricting its use strictly to compatible harness loop workflows.

Related Skills

Explore More Software Engineering Skills→
smith-hornsmith-horn
75

github-code-review

Coordinate parallel AI agents to review GitHub pull requests and enforce quality gates.

Official
Advanced
wangdongzuopinwangdongzuopin
8

sentinel-audit

Audit source-code security and dependencies, then generate Markdown and HTML reports.

Community
Advanced
ryan-nguyen-01ryan-nguyen-01
4

agent-tester

Generates unit, integration, and e2e tests from code diffs and orchestrator context.

Community
Advanced
kurokeitakurokeita
1

find-bugs

Analyze code changes to identify bugs, security vulnerabilities, and quality issues.

Community
Intermediate