Fits whatyou already use.

Point your agent at it.

Add the hosted MCP server and sign in, and your agent queries the graph. Or run the open-source engine yourself.

# hosted: add the MCP server, then sign in with OAuth
$ claude mcp add graphify --transport http https://api.graphify.com/mcp

# Cursor, Codex, Copilot, Windsurf, Cline, Continue, JetBrains, Goose, Zed:
# the same endpoint, OAuth on first use

# or run it yourself with the open-source engine
$ uv tool install graphifyy
$ graphify install

And the rest of it.

How your code comes in, and how CI and tickets reach the graph.

• GitHub and GitLab

  Connect a repository through the GitHub app or GitLab, and Graphify indexes it and reviews its pull requests.

• Jira

  An issue-intent overlay that links tickets to the code they touch, as a heuristic rather than a verdict. Early access.

• Read-scoped API keys

  Keys for CI and scripts, scoped to reading the graph over the HTTP API and nothing else.

Works in 17.

Connect each over MCP and sign in, or wire in the /graphify skill with the open-source engine.

• Claude Code
• Cursor
• GitHub Copilot
• Codex
• Gemini CLI
• Aider
• Amp
• Devin
• Kilo
• OpenCode
• Google Antigravity
• VS Code
• Kiro
• Trae
• Hermes
• Kimi CLI
• Pi

And what else it talks to.

The model backends it can call, what it reads besides code, and where it exports.

Three ways to ask.

The same graph behind all three.

• MCP server

  The hosted server at api.graphify.com/mcp over Streamable HTTP, or a local one from the open-source engine. Read-only graph tools any MCP client can call.

• CLI

  query, path, explain and prs. Plain answers with their citations, straight from the terminal.

• Python package

  Run the open-source engine yourself: one Apache 2.0 package, installed with a single command on your own machine.