Graphify — Codebase to Knowledge Graph
Graphify-Labs/graphify — turns code, docs, SQL schemas, configs, and PDFs into a queryable knowledge graph — no vector store needed. Key differentiator: exact graph relationships instead of embeddings. Weekly trending: +4,023 ⭐, 149,150 total.
How It Works
- Parsing — TypeScript, Python, Go, Rust, SQL, YAML, JSON, Markdown, PDF
- Entity extraction — functions, classes, tables, API endpoints, env vars
- Graph building — nodes = entities, edges = calls, imports, FKs, references
- MCP server — agents (Claude Code, Codex) query graph via tools
Advantages Over RAG
- Precision — no embedding hallucinations
- Relationships — “show all callers of this function” works perfectly
- Updatability — incremental graph updates on code changes
- No vector DB — simpler deploy, fewer resources
How to Start
- Install:
npm install -g @graphify-labs/graphify - In project:
graphify init - Build graph:
graphify build - Run MCP:
graphify mcp - Connect in Claude Code / Codex via MCP config
- “Find all usages of users table”
- “Show call chain from /payment API to DB”
- “What env vars does auth service need?”
- “Any dead code in billing module?”
Example Agent Queries
License: MIT | Language: TypeScript | ⭐ 149,150 (+4,023/week)
