Thanks to visit codestin.com
Credit goes to graphify.com

Skip to content
116.4k
Log inGet started
MCP

Your graph,as native tools.

Connect your coding agent to Graphify over MCP: one endpoint, Streamable HTTP, and an OAuth sign-in on first connect. Your agent queries the graph instead of re-reading files.

Connect

One endpoint, any agent.

Pick your agent, paste what it shows you, and sign in. No key to paste, and every tool is read-only.

Endpoint · Streamable HTTP
https://api.graphify.com/mcp

Choose your agent

Terminal
claude mcp add --transport http graphify https://api.graphify.com/mcp
.mcp.json · shared with the repo
{
  "mcpServers": {
    "graphify": {
      "type": "http",
      "url": "https://api.graphify.com/mcp"
    }
  }
}

Project scope by default; add --scope user to reuse it everywhere. Claude Code runs the sign-in on first use, so there is no key to paste. Check it any time with /mcp.

The tools

Read-only graph tools.

Every one reads the graph. None writes to your code.

  • query_graph

    Search entities and relationships across the graph. Returns the nodes and edges that matter, as context.

  • graphify_node

    One symbol with its body, its edges, and the confidence tag on each.

  • graphify_callers, graphify_callees

    The exact directed callers and callees of a symbol.

  • graphify_trace, shortest_path

    The grounded call path, or the shortest route, between two symbols.

  • graphify_impact

    The blast radius of a change, following every edge out from it.

  • graphify_rank_files

    Rank the files that matter for a question asked in plain English.

Run it yourself

Or serve it locally.

The open-source engine ships the same protocol. Serve your local graph over stdio on your own machine.

# or run the open-source engine's server yourself
$ uv tool install "graphifyy[mcp]"
$ python -m graphify.serve graphify-out/graph.json

Try Graphify now.

Try for free