When should I use this skill instead of standard debugging?

Use this skill for complex failures where an agent is deviating from its defined protocol or tools are being intercepted incorrectly; it is not intended for routine syntax debugging or preventive maintenance.

How does this skill prevent future AI errors?

It 'immunizes' the system by hardening constraints in the SKILL.md file, updating tool signatures, and adding troubleshooting documentation to the component's description.

What system files does this skill analyze during a diagnosis?

The skill analyzes the context log for timestamps, the scratchpad for agent intent, the hooks configuration for interceptions, and the component manifesto for instruction accuracy.

What is the primary purpose of the Repairing State skill?

It is designed to perform forensic investigations into component failures and AI behavioral issues like hallucinations or instruction drift, applying permanent logic fixes rather than temporary patches.

State Repair & Forensic Healing

Name: State Repair & Forensic Healing
Author: Git-Fg

byGit-Fg

•

分析と監視

Diagnoses and permanently resolves AI hallucinations, instruction drift, and component failures through forensic log analysis.

This skill provides a specialized forensic protocol for Claude Code to investigate and fix deep-seated issues within the cattoolkit ecosystem. It moves beyond standard debugging by analyzing 'Black Box' failure data—including context logs, scratchpads, and hook interceptions—to identify exactly why an agent deviated from its instructions. By applying permanent logic updates, constraint hardening, and immunization steps, it ensures that component failures and instruction drifts are not just patched, but structurally prevented from recurring in future operations.

主な機能

01Detection of AI instruction drift and tool usage hallucinations

021 GitHub stars

03Forensic evidence retrieval from context logs and scratchpads

04Component immunization through automated troubleshooting documentation

05Automated resolution of path ambiguity for component definitions

06Hardening of skill constraints to prevent logic recurrence

ユースケース

01Updating outdated skill examples that are incompatible with current toolkit versions

02Fixing an AI agent that consistently uses incorrect or non-existent tool arguments

03Debugging why a specific tool call was blocked by a system hook without explanation

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add git-fg/thecattoolkit repairing-state

For use in Claude.ai and ChatGPT

Download Skill