Wiring your editor¶

whygraph-mcp is a standalone MCP server, so any agent that speaks MCP can use it. whygraph init --agent X writes the right config to the right place for each one.

Run it from the repo you want WhyGraph to analyze:

whygraph init --agent claude

Supported agents¶

Four agents are supported. All of them are project-scoped - the config file is written or merged inside the repo, so you can commit it and every teammate's editor picks it up.

`--agent`	Editor	Config file
`claude`	Claude Code	`.mcp.json` (repo root)
`cursor`	Cursor	`.cursor/mcp.json`
`vscode` (alias `copilot`)	VS Code / GitHub Copilot	`.vscode/mcp.json`
`codex`	OpenAI Codex	`.codex/config.toml`

Run whygraph init --list-agents to print these paths for your own checkout.

The generated config launches whygraph-mcp by bare command name, so the same checked-in file works for everyone who has WhyGraph installed - no absolute paths to scrub.

Claude Code assets¶

--agent claude does one extra thing: it copies a bundled asset tree into .claude/. Re-running leaves your existing files alone; pass --force to overwrite them.

whygraph init --agent claude --no-install-assets   # MCP wiring only, skip the .claude/ copy
whygraph init --agent claude --force               # overwrite existing .claude/ files

Useful flags¶

Flag	What it does
`--print`	Print the MCP snippet to stdout instead of writing any file. Good for pasting by hand.
`--list-agents`	List supported agents and their config paths, then exit.
`--install-assets / --no-install-assets`	Copy (or skip) the agent's bundled assets. Default: copy. No-op for agents with no asset tree.
`--skip-preflight`	Skip the host-tool diagnostics. For known-good scripted environments.
`--force`	Overwrite existing asset files in the destination directory.

Verify¶

After wiring, confirm the server launches:

whygraph-mcp   # Ctrl-C to exit

If it starts cleanly, your editor can start it too. Next, see how an agent actually calls the tools.