Cross-language code intelligence for AI agents. Builds a semantic graph of functions, classes, imports, and call chains across 38 languages and exposes it through MCP tools – so you query structure instead of grepping files.
Cross-language code intelligence for AI agents. Builds a semantic graph of functions, classes, imports, and call chains across 38 languages and exposes it through MCP tools - so you query structure instead of grepping files.This is the skill's description: the text your AI reads to decide when the skill applies.
codegraph/SKILL.md
---
name: CodeGraph
description: Understand a codebase as a graph, not a pile of files. When you need structure - who calls this, what does it import, what breaks if I change it - query CodeGraph's MCP tools instead of grepping, and trust the structured result.
---
# CodeGraph
CodeGraph gives you structured code understanding through a semantic graph of
the codebase - functions, classes, imports, and call chains across 38 languages -
served over MCP. Use it whenever a question is about *structure* rather than
text.
## When to reach for it
Prefer CodeGraph over raw `grep`/file reads when the user asks:
- "Who calls this function?" / "What does this depend on?"
- "What's the blast radius if I change X?" - impact and call-graph traversal.
- "Where is this concept implemented?" - semantic search over full-body
embeddings, not just string matches.
- "Why did we do it this way?" - the persistent memory layer surfaces past
debugging insights across sessions.
## How to use it well
… load the full skill through Skill Me