What is a context budget in a Claude Skill?

Skills have a character limit (typically ~15,000) for their definitions. This skill helps you keep your core instructions concise and move large examples to secondary files to avoid truncation.

How does this skill improve Claude's ability to find my tools?

It checks your descriptions for 'trigger phrases' like 'Use when user asks...' which are essential for Claude to know exactly when a specific skill is relevant to the conversation.

Can it help secure my custom Claude skills?

Yes, it identifies unrestricted tool access—like generic Bash permissions—and suggests scoped patterns like Bash(git:*) to minimize security risks.

What does the enhance-skills plugin actually do?

It analyzes the structure of Claude Code skills (SKILL.md files) to ensure they follow best practices for discoverability, security, and model performance.

Skill Quality & Security Optimizer

Name: Skill Quality & Security Optimizer
Author: agent-sh

byagent-sh

•

開発者ツール

Analyzes and validates Claude Code SKILL.md files to ensure high trigger quality, proper tool scoping, and optimal discoverability.

This meta-skill acts as a linter and optimizer for the Claude Code ecosystem, specifically designed to audit and improve the definitions of other skills. It scans SKILL.md files to verify frontmatter compliance, ensure effective trigger phrases are present for auto-discovery, and enforce critical security patterns like scoped tool access. By identifying anti-patterns and providing auto-fixes for structural issues, it helps developers build reliable, secure, and highly discoverable AI capabilities that stay within the model's context budget.

主な機能

011 GitHub stars

02Security auditing for tool restriction patterns and unrestricted Bash access

03Automated validation of SKILL.md frontmatter and YAML schema

04Trigger quality analysis to improve natural language auto-discovery

05Context budget monitoring to prevent skill truncation and efficiency loss

06High-certainty auto-fixing for metadata and common structural errors

ユースケース

01Optimizing skill descriptions to ensure they trigger correctly during AI sessions

02Maintaining large-scale skill directories by offloading complex content to references

03Auditing a repository of custom Claude skills for security and schema consistency

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add agent-sh/enhance enhance-skills

For use in Claude.ai and ChatGPT

Download Skill