codegraph is a local-first code context engine and MCP server that builds a persistent knowledge graph of your source repositories in SQLite. It gives AI coding assistants deep structural awareness — symbols, call graphs, dependencies, and semantic search — without sending a single byte to the cloud.
Single binary. Zero config. No external databases. No API keys.
Why codegraph?
AI coding assistants are powerful, but they spend most of their token budget discovering what to change — grepping files, reading code, reconstructing call graphs from partial evidence.
codegraph shifts that cost. One call to context_for_task returns the exact files, symbols, and relationships an agent needs. One call to agentic_query gets a synthesized answer backed by graph traversal and semantic search.
❌ Without codegraph
AI reads files one by one → greps for patterns → burns tokens on context-gathering
✅ With codegraph
AI calls context_for_task("add retry logic to HTTP client")
→ instantly gets relevant files, functions, callers, callees, and tests
How It Works
Your Code ──▶ tree-sitter AST ──▶ SQLite Graph ──▶ MCP Tools ──▶ AI Assistant
│ │ │
12 languages symbols, edges 29 tools
framework detect embeddings agentic reasoning
import resolution session memory hybrid search
codegraph index . walks your repo, parses every file with tree-sitter, resolves imports using four strategies (exact, name, suffix, method-receiver), and writes a fully-linked symbol graph into a local codegraph.sqlite. The MCP server then exposes that graph to any compatible AI assistant via 29 structured tools — no cloud, no Docker, no API keys.
Features
🔍 Parsing & Indexing
- Tree-sitter parsing for all 12 supported languages — robust AST extraction, not regex
- 4-strategy import resolution — exact, name, suffix, and method-receiver matching
- Cross-language linking — connects symbols across language boundaries
- Incremental updates — only re-indexes changed files; fast on large repos
- Framework detection — recognizes 20+ frameworks (Express, Django, gin, React, Spring, Laravel, …)
🔎 Search & Query
- Hybrid search — vector similarity (Ollama embeddings) + FTS5, fused with Reciprocal Rank Fusion
- Semantic search — find code by meaning, not just text
- Call graph traversal — callers, callees, transitive dependency chains
- Impact analysis — know what breaks before you change it
- Dead code detection — find symbols with zero references
- Architecture overview — language breakdown, entry points, hub symbols, coupling metrics
🤖 AI Integration
- 29 MCP tools — comprehensive API for AI coding assistants
- Agentic reasoning — ReAct loop over a local Ollama LLM that chains tools and synthesizes answers
- Context building — one tool call returns everything an agent needs for a task
- Session memory — persist reads, edits, decisions, and facts across sessions
- Token benchmarking — measure savings vs. naive file reading
📊 Graph Analytics
- PageRank — find the most important symbols in your codebase
- Coupling metrics — identify tightly coupled file pairs
- Cycle detection — find circular dependencies at the file level
- Interactive visualization — D3.js force-directed graph with search and zoom
🛠 Developer Experience
- Single Go binary — no runtime dependencies, cross-platform
- Zero-config SQLite — no Docker, no external databases
codegraph install— auto-detects and configures Claude Code, Cursor, Windsurf, Gemini CLI- File watching — automatic re-indexing on changes
- 100% local — no data leaves your machine
Supported Languages
All languages use tree-sitter for AST parsing:
Node.js repos are supported; full tree-sitter node support is still in progress.
Quick Start
1. Install
go install github.com/isink17/codegraph/cmd/codegraph@latest
Requires Go 1.23+ and a C compiler (for tree-sitter CGo bindings).
Build from source
git clone https://github.com/isink17/codegraph
cd codegraph
go build ./cmd/codegraph
go test ./...
Requires Go 1.23+ and a C compiler (for tree-sitter CGo bindings).
Clean rebuild
cd your-project
codegraph index . --rebuild
Use this after parser or indexer changes when you need a true full reindex from scratch.
codegraph index . --rebuild needs exclusive access to the repo database.
If rebuild fails because the DB is in use, stop codegraph serve or other codegraph processes and retry.
Use codegraph clean . for database maintenance tasks like WAL checkpointing, VACUUM, FTS optimize, ANALYZE, and incremental vacuum.
Version
codegraph --version
codegraph version
Prints the installed local version only. It does not contact GitHub.
Normal commands check GitHub Releases at most once every 24 hours. Set
DO_NOT_TRACK or CODEGRAPH_NO_UPDATE_CHECK to a non-empty value other than
0 or false to disable the check. This also prevents creation or update of
the local version-check state file.
2. Auto-configure your AI tool
codegraph install
Detects Claude Code, Cursor, Windsurf, and Gemini CLI and writes the MCP config automatically.
3. Index your project
cd your-project
codegraph index .
4. Start the MCP server
codegraph serve
Repo root is auto-detected from git (or falls back to your current working directory).
That's it. Your AI assistant now has deep structural code understanding.
MCP Setup
Auto-configure (recommended)
codegraph install
Manual Setup
{
"mcpServers": {
"codegraph": {
"command": "codegraph",
"args": ["serve"]
}
}
}
{
"mcpServers": {
"codegraph": {
"command": "codegraph",
"args": ["serve"]
}
}
}
[mcp_servers.codegraph]
command = "codegraph"
args = ["serve"]
startup_timeout_sec = 60
See the examples/ directory for more configuration samples.
MCP Tools (29)
Code Intelligence
Architecture & Analysis
Repository Management
Session Memory
Agentic
CLI Reference
# Setup
codegraph install # Auto-configure AI tools
codegraph doctor # Check installation health
codegraph config show # Show current config
codegraph --version # Print current version
codegraph version # Print current version
# Indexing
codegraph index <path> # Full index
codegraph update <path> # Incremental update
codegraph watch <path> # Watch and auto-reindex
codegraph clean <path> # Clean database
# MCP Server
codegraph serve [--repo-root <path>] # Start MCP server (auto-detects repo root)
# Query
codegraph stats <path> # Graph statistics
codegraph find-symbol <path> <query> # Find symbols
codegraph search <path> <query> # Full-text symbol search
codegraph callers <path> --symbol <name> # Find callers
codegraph callees <path> --symbol <name> # Find callees
codegraph impact <path> --symbol <name> # Impact analysis
# Testing
codegraph affected-tests [--stdin] <files> # Tests affected by changed files
# Visualization & Export
codegraph visualize [--repo-root <path>] # Interactive D3.js graph (auto-detects repo root)
codegraph graph export <path> --format dot # Export as Graphviz DOT
codegraph graph export <path> --format json # Export as JSON
# Benchmarking
codegraph benchmark # Token savings benchmark
Affected Tests with Git
# Find tests affected by uncommitted changes
git diff --name-only | codegraph affected-tests --stdin
# CI integration
TESTS=$(git diff --name-only HEAD~1 | codegraph affected-tests --stdin)
go test $TESTS
Replace --repo-root . with nothing if you are already in the repo you want to inspect.
Optional: Embeddings & Agentic Mode
codegraph works fully without any external services. Enable these for enhanced capabilities:
Vector Embeddings (via Ollama)
Enables hybrid semantic search (vector + FTS):
# Install and pull embedding model
ollama pull nomic-embed-text
# Initialize repo config
codegraph config init --repo .
Edit .codegraph/config.json:
{
"embedding": {
"enabled": true,
"model": "nomic-embed-text"
}
}
Then re-index to generate embeddings:
codegraph index . --force
Agentic Reasoning (via Ollama)
The agentic_query tool uses a local LLM to reason over the graph with a ReAct loop:
ollama pull llama3.2
Edit .codegraph/config.json:
{
"agent": {
"enabled": true,
"model": "llama3.2"
}
}
Configuration
Global config
Repo config
Created with codegraph config init --repo . at .codegraph/config.json:
{
"include": [],
"exclude": ["vendor/**", "node_modules/**"],
"languages": [],
"embedding": {
"enabled": false,
"model": "nomic-embed-text"
},
"agent": {
"enabled": false,
"model": "llama3.2"
}
}
Repo Root Resolution
When --repo-root (CLI) or repo_root (MCP tool parameter) is omitted, codegraph resolves the repo root using:
- Per-call
repo_rootMCP tool parameter --repo-rootCLI flag (process-level default)git rev-parse --show-toplevelfrom the current working directoryos.Getwd()(current working directory)- Return error
Ignore file
Create .codegraphignore in the repo root (same syntax as .gitignore):
build/
dist/
*.generated.go
Note: Common generated directories (
node_modules,.next,.nuxt, etc.) are always skipped and cannot be un-ignored.
Architecture
cmd/codegraph CLI entrypoint
internal/
agent/ Agentic reasoning (ReAct loop over Ollama)
cli/ Command handlers and MCP auto-configuration
config/ Config loading and path resolution
embedding/ Vector embedding (Ollama HTTP client)
export/ JSON and DOT graph export
framework/ Framework detection (20+ frameworks)
graph/ Core types (Symbol, Edge, Reference, …)
indexer/ Repository scan, incremental updates, embedding
mcp/ MCP stdio server (29 tools)
parser/ Parser interface and adapters
treesitter/ Tree-sitter adapters (12 languages)
golang/ Go AST parser (legacy)
python/ Python heuristic parser (legacy)
heuristic/ Regex-based parsers (legacy fallback)
query/ Query orchestration and hybrid search
store/ SQLite storage, migrations, graph analytics
viz/ Interactive D3.js graph visualization
watcher/ File watch and debounced updates
Building from Source
git clone https://github.com/isink17/codegraph
cd codegraph
go build ./cmd/codegraph
go test ./...
Requires Go 1.23+ and a C compiler for the tree-sitter CGo bindings.
For a clean rebuild after parser/indexer changes:
codegraph index . --rebuild
This rebuild path needs exclusive access to the repo database. If it fails because another codegraph process is holding the DB, stop that process and retry.
License
This project is licensed under the Functional Source License, Version 1.1, MIT Future License (FSL-1.1-MIT).
On the second anniversary of each version's release, that version converts to the MIT License. See LICENSE for full terms.