repair-agent

Audit Claude Code agents across seven dimensions and apply critical fixes.

270|16|Updated Feb 3, 2026
One-click install
npx skills add https://github.com/gupsammy/Claudest --skill repair-agent
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: repair-agent
Source: https://github.com/gupsammy/Claudest/tree/main/plugins/claude-skills/skills/repair-agent
Command: npx skills add https://github.com/gupsammy/Claudest --skill repair-agent

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill diagnoses and rectifies issues in existing Claude Code agents, ensuring they meet a high standard of quality and functionality.

Core Features & Use Cases

  • Comprehensive Auditing: Evaluates agents across 7 dimensions, including description quality, frontmatter, system prompt voice, agentic vs. deterministic split, efficiency, process completeness, and anatomy.
  • Gap Identification: Pinpoints missing elements or areas for improvement beyond just fixing violations.
  • Structured Improvement Plan: Generates a detailed report outlining violations, gaps, and required changes.
  • Automated Application of Fixes: Applies critical and major improvements directly to the agent file.
  • Use Case: You have an agent that isn't triggering correctly or is performing inconsistently. Use this Skill to identify the root cause and get a clear plan for fixing it.

Quick Start

Use the repair-agent skill to audit the agent file located at agents/my-buggy-agent.md.

Frequently Asked Questions about repair-agent

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

FAQPage Schema
How do I audit and improve a buggy Claude Code agent?

To audit and improve a Claude Code agent, the repair-agent skill evaluates the agent file against a gold standard across seven dimensions, identifying violations and gaps, generating a structured improvement report, and applying critical and major fixes directly to the file.

What is the best way to fix a Claude Code agent that isn't triggering correctly?

Fixing a Claude Code agent that isn't triggering correctly requires a comprehensive audit of its description quality, frontmatter, and system prompt voice to pinpoint inconsistencies, identify root causes, and generate a structured improvement plan for resolving the issue.

How does agent auditing work across the seven quality dimensions?

Agent auditing works by evaluating the agent file across seven dimensions: description quality, frontmatter, system prompt voice, agentic vs. deterministic split, efficiency, process completeness, and anatomy, ensuring the agent meets a high standard of quality and functionality.

Do I need reference files to audit a Claude Code agent?

Yes, you need reference files for agent anatomy and frontmatter definitions to perform a comprehensive analysis, as the audit requires these files to evaluate the agent against a gold standard and identify violations and gaps accurately.

Can I automatically apply fixes to my agent file after an audit?

Yes, you can automatically apply critical and major improvements directly to the agent file after the audit generates a detailed improvement report outlining the violations, gaps, and required changes needed to rectify the issues.

Why does my Claude Code agent perform inconsistently and how can I identify gaps?

A Claude Code agent performs inconsistently due to gaps and violations in its anatomy or prompt structure, which you can identify by running a comprehensive audit that pinpoints missing elements and generates a structured improvement plan.