@ttsc/graph

Gives your AI coding agent a map of your TypeScript codebase, over MCP, so it answers "how does this work?" without opening file after file.

Ask an agent like Claude Code or Codex about your project and it works file by file: open one, follow an import, open the next, until it has pieced the picture together by hand. Slow, token-hungry, and every relationship is a guess from reading text.

@ttsc/graph hands it the map up front: what calls what, what depends on what, where each piece lives. Drawn by the real TypeScript compiler, so it is exact, not skimmed.

On a public benchmark, an agent answered while reading zero files, cutting tokens by 77% to 86% and tool calls by 94% to 95% (see the benchmark).

You can also browse the whole map. This is TypeORM in 3D, colored by kind (live viewer):

Setup

Install

npm install -D ttsc @ttsc/graph typescript@rc

@ttsc/graph reads the map from the program ttsc already type-checked, so install the two together.

Connect your agent

Add the server to your agent's MCP config, once.

For Claude Code, that is a .mcp.json in your project root:

{
  "mcpServers": {
    "ttsc-graph": {
      "command": "npx",
      "args": ["-y", "@ttsc/graph"]
    }
  }
}

Start your agent from your project root so the server finds your tsconfig.json. The agent queries the map on its own; you never call it by hand.

Browse it in 3D

Run this in your own project to open that 3D map in your browser, served from a local port:

npx @ttsc/graph view

Sponsors

Thanks for your support.

Your donation encourages ttsc development.

References

@ttsc/graph is inspired by codegraph, which first put a code graph in front of an agent over MCP. The benchmark here is a faithful port of codegraph's.

The difference is where the map comes from. codegraph parses the shape of your code and infers how the pieces connect, while @ttsc/graph asks the real TypeScript compiler, which has already resolved every import and reference, so the map is exact rather than inferred.