Can this skill find broken links in my documentation?

Yes, it extracts file paths and npm commands from your memory files and validates them against your actual project structure to identify broken references.

Why does content positioning matter in project memory?

LLMs often suffer from a 'lost in the middle' effect where they recall information at the start and end of a file better than the middle. This skill ensures your most critical rules are placed in high-attention zones.

What is the difference between CLAUDE.md and AGENTS.md?

CLAUDE.md is the standard memory file for Claude Code, while AGENTS.md is a cross-tool format used by OpenCode, Codex, and other AI assistants. This skill helps you maintain both for maximum flexibility.

How does it improve AI instruction following?

It identifies 'weak' language (like 'try to' or 'should') and negative constraints, suggesting 'strong' positive alternatives (like 'MUST' or 'ALWAYS') that research shows LLMs follow more reliably.

Enhance Claude Memory

Name: Enhance Claude Memory
Author: agent-sh

byagent-sh

•

生产力与工作流

Optimizes AI project memory files like CLAUDE.md and AGENTS.md to improve instruction following and agent performance.

Enhance Claude Memory is a specialized tool designed to audit and refine AI context files, ensuring they provide the most effective guidance for Claude and other AI agents. By analyzing instruction hierarchy, validating file references, and applying prompt engineering best practices—such as positive reinforcement and optimal content positioning—this skill transforms messy documentation into high-performance project memory. It helps developers reduce token waste and minimize AI errors by ensuring critical rules are placed where models can best 'see' them.

主要功能

01Verifies file and command references against the local filesystem

02Calculates token metrics and identifies README content duplication

03Detects 'lost in the middle' positioning issues for critical project rules

04Validates structure and hierarchy for CLAUDE.md and AGENTS.md files

05Analyzes instruction effectiveness based on prompt engineering research

061 GitHub stars

使用场景

01Standardizing documentation for cross-tool compatibility between Claude and OpenCode

02Auditing existing project memory files for better AI instruction compliance

03Optimizing large context files to reduce token usage and improve response speed

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add agent-sh/enhance enhance-claude-memory

For use in Claude.ai and ChatGPT

Download Skill