← Back to research
•·4 min read·company

CodeGraphContext

CodeGraphContext — MCP server that indexes local code into a graph database for AI assistants. The MIT-licensed alternative to GitNexus. 3.7k stars, Python, 22 languages, ~31k PyPI downloads/month, multiple graph backends.

https://cgc.codes ↗
coding-agentscode-intelligenceknowledge-graphmcpopen-sourcepythongithub

Key takeaways

  • MCP server + CLI that indexes local code into a graph database, providing structured context to AI assistants. The MIT-licensed alternative to GitNexus
  • 3.7k stars, 694 forks, ~31k PyPI downloads/month as of June 2026. Python. Supports 22 languages including Python, TypeScript, Java, Go, Rust, and C#
  • Pluggable graph backends — FalkorDB Lite (default), KuzuDB, Neo4j, LadybugDB, and Nornic DB — so teams are not locked into one database engine
  • Actively developed (commits as of June 11, 2026): v0.4.7 stable, VS Code extension in alpha, SCIP indexing for C/C#, and shareable pre-indexed .cgc bundles. MIT license makes it safe for commercial and enterprise use

FAQ

What is CodeGraphContext?

An MCP server and CLI tool that indexes local code into a graph database, providing structured context (symbols, dependencies, call chains) to AI coding assistants like Claude Code and Cursor.

How much does CodeGraphContext cost?

It is free and open source under the MIT license, installed from PyPI. There is no paid tier — you supply your own graph database backend, with embedded options (FalkorDB Lite, KuzuDB) that require no separate server.

How does it compare to GitNexus?

GitNexus has deeper Claude Code integration (hooks, skills) and a larger star count, but uses a noncommercial license. CodeGraphContext is MIT-licensed and supports multiple standard graph backends — making it safer for commercial use.

Is CodeGraphContext still maintained?

Yes. As of June 11, 2026 the repo had commits pushed that same day, a stable v0.4.7 release from May 2026, a v0.5.0 in development, and an alpha VS Code extension.

Overview

CodeGraphContext is an MCP server and CLI tool that indexes local code into a graph database to provide structured context to AI assistants. It bridges the gap between deep code graphs and AI context — giving agents symbol-level understanding of codebases.

Status (verified June 11, 2026): Actively developed. The GitHub repo had commits pushed on June 11, 2026, with 3,702 stars (up from ~2.2k in March), 694 forks, and 185 open issues. Latest stable release is v0.4.7 (May 7, 2026), with v0.5.0 in development and a VS Code extension released in alpha in May 2026.

Key stats: 3.7k stars, MIT license, Python. Created August 2025. ~31k PyPI downloads in the past month as of June 2026.

What It Does

CodeGraphContext parses a codebase into a queryable knowledge graph — symbols, dependencies, and call chains — and serves it to AI assistants over MCP. It supports 22 languages, including Python, JavaScript, TypeScript, Java, C, C++, C#, Go, Rust, Ruby, PHP, Swift, and Kotlin.

Recent additions:

  • Pluggable graph backends — FalkorDB Lite (default on Unix), KuzuDB (cross-platform), Neo4j, LadybugDB, and Nornic DB
  • SCIP indexing — optional precise indexing for C and C#
  • Pre-indexed bundles — shareable .cgc files so teams can distribute an indexed graph without re-indexing
  • Interactive visualization — generates standalone HTML graph views that run in any browser
  • VS Code extension — v0.1.0 alpha shipped May 2026

Pricing

Free and open source under the MIT license; installed from PyPI. There is no commercial tier as of June 2026 — the embedded database defaults (FalkorDB Lite, KuzuDB) mean no external server cost either.

Competitive Position

Strengths: MIT license (enterprise-safe, unlike GitNexus's noncommercial license). Multiple standard graph backends (no single-engine lock-in). Broad 22-language support. Sustained release cadence and growing adoption (~31k PyPI downloads/month).

Weaknesses: Less deep editor integration than GitNexus (no Claude Code hooks or skills; VS Code extension still alpha). Smaller star count (3.7k vs GitNexus's 14k). Less visual tooling.

Cautions

  • Community-stage maturity — 185 open issues against a small watcher base; quality varies across the 22 supported languages, many added via community contribution.
  • Thin independent validation — adoption signal is downloads and stars, not public production case studies. Earlier "80% time savings" claims are project-reported, not independently verified.
  • Fast-moving surface — backend options and CLI behavior have changed across recent releases (v0.4.x); pin versions in team setups.

What Developers Say

Public developer discussion is minimal as of June 11, 2026. The February 2026 Show HN launch drew 2 points and zero comments, and no substantive Reddit threads or independent reviews were found. The absence is itself a signal: download growth (~31k/month on PyPI) is outpacing public conversation, suggesting quiet utility-tool adoption rather than community buzz.

Bottom Line

CodeGraphContext is the pragmatic, permissively-licensed entry in the code-graph MCP space. It is verifiably alive — daily commits, a May 2026 stable release, and roughly 31k PyPI downloads a month — and its MIT license plus pluggable standard backends make it the safe default where GitNexus's noncommercial license is a blocker.

Recommended for: Teams that want graph-structured code context for AI agents with a commercial-safe license, and polyglot codebases that benefit from 22-language coverage.

Not recommended for: Anyone needing deep editor automation (hooks, skills) today — GitNexus is ahead there — or teams that require battle-tested, case-study-backed tooling.

Outlook: The multi-backend strategy and VS Code extension point toward becoming infrastructure rather than a single tool. The open question is whether it converts download growth into a visible community before editor-native codebase indexing catches up.

Research by Ry Walker Research • methodology

Sources

Share on XShare on LinkedIn