1. Conduid
  2. Data
  3. GraphRepo
MCP server · Data

GraphRepo

Parse any codebase into a Neo4j graph, visualize it with D3, and query it through Claude via MCP.

Unclaimed data
34Low

Scored 4 months ago · breakdown

About GraphRepo

GraphRepo is an MCP server in the Data category: parse any codebase into a Neo4j graph, visualize it with D3, and query it through Claude via MCP. It has been installed 0 times through Conduid.

Install

Clone
git clone https://github.com/the-muses-ltd/GraphRepo

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Ask AI

Ask AI about GraphRepo

Powered by Claude · Grounded in docs

I know everything about GraphRepo. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

  • ·README presentNot checked yet.
  • ·License declaredNot checked yet.
  • ·Tests presentNot checked yet.
  • ·Dependencies pinnedNot checked yet.
  • ·No dynamic code executionNot checked yet.
  • ·Scoped permissionsNot checked yet.

README

GraphRepo

A knowledge graph for your codebase. GraphRepo parses your repository into an in-memory graph that captures every file, function, class, import, and call relationship — then exposes it to AI code assistants via MCP so they can traverse the full interconnected structure of your code, not just individual files.

TypeScript Python C Swift MCP VS Code Marketplace

GraphRepo VS Code Extension

Why

AI assistants read files one at a time. They can't see how your code connects — which functions call which, how modules depend on each other, or where a change will ripple through. GraphRepo gives them a queryable map of your entire codebase so they can reason about architecture, trace dependencies, and understand context that spans hundreds of files.

What it does

  • Parses your repo using Tree-sitter AST analysis (TypeScript, JavaScript, Python, C, C++, C#, Swift) and indexes all file types
  • Builds an in-memory knowledge graph of files, functions, classes, imports, call chains, and folder structure
  • Detects communities using Louvain algorithm to identify code modules and clusters
  • Generates embeddings using Transformers.js for semantic search (local, no API keys)
  • Serves an MCP with 10 tools that let Claude (or any MCP-compatible assistant) traverse your code graph — search by name, trace call chains, walk dependency trees, and run semantic queries
  • Visualizes the graph in a VS Code sidebar with an interactive D3-force layout
  • Tracks your editor — the graph follows your cursor and highlights the node you're working on
  • Shows MCP activity — watch the graph light up in real time as your AI assistant queries it
  • Supports multiple repos — parse as many projects as you want, each scoped by name
  • Zero external dependencies — no Docker, no database servers, everything runs locally

Getting Started

1. Install

From the VS Code Marketplace (recommended):

Search for "GraphRepo" in the Extensions panel (Ctrl+Shift+X), or install from the command line:

code --install-extension the-muses-ltd.graphrepo

From a VSIX file:

Download the latest .vsix from GitHub Releases, then:

code --install-extension graphrepo-0.7.0.vsix

2. Parse your repo

Open any project in VS Code, then:

  • Press Ctrl+Shift+PGraphRepo: Parse Workspace

The graph view in the sidebar will populate automatically. Graph data is stored in .graphrepo/ inside your workspace.

3. Connect your AI assistant

MCP is auto-configured. After parsing, GraphRepo creates a .mcp.json in your workspace root so Claude Code picks up the MCP tools automatically. Just restart Claude Code after the first parse.

You can also run Ctrl+Shift+PGraphRepo: Configure MCP for Claude at any time.

For other MCP clients, add this to your config manually:

{
  "mcpServers": {
    "graphrepo": {
      "command": "node",
      "args": ["<extension-path>/dist/mcp-server.cjs", "serve"],
      "env": {
        "GRAPHREPO_DATA_FILE": "<workspace>/.graphrepo/graph.json",
        "GRAPHREPO_REPO_PATH": "<workspace>"
      }
    }
  }
}

Once connected, your AI assistant can traverse your code graph — tracing call chains, walking dependency trees, and understanding relationships that would be invisible from reading files alone.

MCP Tools

Tool Description
search_code Full-text search for functions, classes, and variables by name
get_file_structure List all entities defined in a file
get_dependencies What does this file/module import?
get_dependents What imports this file/module?
get_call_graph Trace function call chains (callers, callees, or both)
find_related N-hop graph exploration from any node
run_traversal Structured graph traversal with edge type and direction filters
get_summary Repo-wide statistics (files, functions, languages, etc.)
get_communities Code community detection results at different hierarchy levels
semantic_search Find code by meaning using local embeddings

Graph Schema

Nodes: File, Function, Class, Interface, Variable, Module, Folder, Community

Relationships: CONTAINS, IMPORTS, IMPORTS_EXTERNAL, CALLS, HAS_METHOD, EXTENDS, CONTAINS_FILE, CONTAINS_FOLDER, BELONGS_TO_COMMUNITY, PARENT_COMMUNITY

VS Code Extension

GraphRepo includes a VS Code extension with a sidebar graph view:

  • Follow editor — graph automatically centers on the file/function you're editing
  • Click to navigate — click any node to open it in your editor
  • Follow MCP — watch the graph highlight nodes as your AI assistant queries them
  • Community coloring — see how your code clusters into modules
  • All toggleable from the sidebar controls

Commands

Command Description
npm run parse -- <path> Parse a repo into the graph
npm run parse -- <path> --clear Clear entire graph, then parse
npm run serve Start MCP server (STDIO transport)
npm run viz Start visualization server on :3000
npm run build:ext Bundle extension
npm run build:mcp Bundle standalone MCP server
npm run build:vscode Build extension + webview
npm run typecheck Run TypeScript type checking

Architecture

src/
├── parser/          # Tree-sitter AST parsing (TS, JS, Python, C, C++, C#, Swift)
│   └── languages/   # Per-language extraction logic
├── graph/           # Graphology in-memory graph (store, sync, queries, persistence)
├── graphrag/        # Community detection (Louvain) + Transformers.js embeddings
├── mcp/             # MCP server with 10 tools
├── extension/       # VS Code extension + webview
├── web/             # D3-force visualization (dark theme)
└── cli/             # Commander CLI entry point

Tech Stack

TypeScript, web-tree-sitter, graphology, @huggingface/transformers, @modelcontextprotocol/sdk, d3-force, esbuild, Commander

README mirrored from the source repository 4 months ago. The original is authoritative.

Questions

About GraphRepo

How do I install GraphRepo?

Run git clone https://github.com/the-muses-ltd/GraphRepo, then add the server to your MCP client's configuration. Conduid has recorded 0 installs, so the command is known to work with current clients.

Is GraphRepo safe to use with an AI agent?

Its trust score is 34 out of 100 (low). Conduid hasn't run static security checks on this repository yet, so review the source yourself before granting it credentials. It has no ConduID identity yet, so agent calls to it are not receipted.

Is GraphRepo still maintained?

Conduid hasn't recorded a commit date for this repository yet. Check the repository directly for recent activity.