A Claude Code skill that teaches Claude how to install, configure, and use the code-memory MCP server for semantic code intelligence inside any Git repository.

When invoked, this skill guides Claude to:

• Index a local Git repository into a SQLite-backed semantic search database
• Run AST-aware symbol lookups, topic discovery, and reference searches
• Search README/docs/docstrings with natural-language queries
• Dig into Git history, file history, and blame data
• Identify dead-code candidates before a cleanup

Drop SKILL.md into your Claude Code skills directory:

# Personal skills (Claude Code)
~/.claude/skills/code-memory/SKILL.md

# Or OpenClaw workspace
~/.openclaw/workspace/skills/code-memory/SKILL.md

Claude will automatically offer to use the skill whenever you start work in a Git repository where semantic search would help.

• Python >= 3.13
• uv / uvx (recommended)
• ~600 MB disk for the default embedding model (jinaai/jina-code-embeddings-0.5b)
• ~1 GB RAM when the model is loaded

All credit for the code-memory MCP server goes to its original author:

Kapil Lamba — github.com/kapillamba4

Project repository: github.com/kapillamba4/code-memory

This repository contains only a Claude Code skill wrapper (SKILL.md) that documents how to install, configure, and use that tool. It does not contain or redistribute any of the original source code.

The skill was reviewed against code-memory version v1.0.32 / commit 563788a2a7d015699f20251d404aeb293346f40c.

The skill file in this repository is released under CC0 1.0 Universal (public domain). The upstream code-memory project has its own separate license.