Skip to main content

OpenClaw MCP: how to connect Graphify

By Steven Van ·

The ClawHub plugin links OpenClaw to repositories already indexed in Graphify Cloud, using browser sign-in and no API key.

Graphify has published a guide to connecting OpenClaw to its hosted MCP server. The route is a plugin, @graphify-labs/graphify-mcp, listed as Graphify MCP on ClawHub. It connects OpenClaw to repositories already indexed in Graphify Cloud, through https://api.graphify.com/mcp. Sign-in happens in the browser, with no API key to paste. OpenClaw does not build the graph on the user's computer.

The steps are in Graphify's guide, and cover Graphify setup in OpenClaw.

What you need first

  • OpenClaw with remote MCP, OAuth and Claude-format bundle support. The package was prepared for OpenClaw 2026.9.8, and older releases may differ.
  • A model enabled in OpenClaw with permission to call tools. This is separate from the Graphify plan.
  • A repository connected in Graphify Cloud, with its status showing Indexed on Repositories and builds.

Install and sign in

Install with openclaw plugins install clawhub:@graphify-labs/graphify-mcp, then check it with openclaw plugins inspect graphify-mcp. Anyone who has cloned the graphify-mcp repository can install the same bundle from integrations/openclaw. If the install asks to accept capabilities, review the Graphify MCP server and accept it. Noninteractive installs need --accept-capabilities. Installing does not sign you in.

To connect, register the server and log in:

  • openclaw mcp set graphify '{"url":"https://api.graphify.com/mcp","transport":"streamable-http","auth":"oauth"}'
  • openclaw mcp login graphify
  • openclaw mcp doctor graphify --probe

The same connection can be merged into the OpenClaw configuration under mcp.servers.graphify. If a server named graphify already exists, run openclaw mcp show graphify first. Keep a single connection with that name, since a second name for the same URL shows the tools twice. No password or token goes in the configuration, because OpenClaw stores the OAuth credentials. Start a new agent session afterwards.

First questions to ask

Ask OpenClaw to list your Graphify workspaces and repositories. Then pick one and ask it to find the function that handles authentication, and to show that function's callers and linked tests with file paths as evidence. A proper connection check includes a repository lookup, because tool discovery alone does not show the index is reachable.

Answers are static and do not run the code or tests, so the guide advises opening the cited file and line before changing code. Some tools save a repository note or change the selected workspace, so each tool request should be read before it is approved.

If it does not work

  • Graphify tools hidden: check the plugin is enabled and the tool policy allows it. A bundle-mcp denial hides them.
  • Sign-in fails: run the doctor probe, then log in again.
  • No repositories or empty answers: the repository is not indexed yet or is in a different workspace.

The local command graphify install --platform claw installs a skill for a graph on your own machine and does not install the ClawHub plugin. The PyPI package for the local engine is graphifyy, with two y's.

Graphify
Graphify
Turn a codebase into a knowledge graph your AI assistant can trace — answers come back as graph paths with real file:line citations.
View Graphify →
Sources

Original source

Read OpenClaw MCP: how to connect Graphify on Creators Toolbox