Client Setup
Connect statshawk from Claude, ChatGPT, Grok, Gemini, Cursor, or any MCP client.
One URL, whatever your client calls it.
statshawk speaks the standard MCP Streamable HTTP transport, so any client that supports
MCP can connect to https://mcp.statshawk.ai/mcp. Every platform names the thing you're
adding differently: a custom connector in Claude, an app in ChatGPT, a connector
in Grok, a connected app in Gemini. The URL is the only stable contract.
One click installs the MCP server into Cursor and opens the OAuth flow on first use. Requires Cursor 0.45 or later.
install in cursor// or open settings → mcp → add new mcp server → url
$ https://mcp.statshawk.ai/mcpClaude Code (CLI)
One command, OAuth handles the rest.
Claude (web, desktop, and mobile)
Available on every plan (Free is limited to one connector). Open Customize → Connectors → Add custom connector, name it statshawk, paste the URL, and click Connect. OAuth handles sign-in. On Team and Enterprise, an Owner or Primary Owner adds it under Organization settings → Connectors.
ChatGPT
ChatGPT calls these apps (renamed from connectors). Pro accounts get read tools via Settings → Apps → Advanced settings → Developer mode, then create a new app with the URL. On Business, Enterprise, and Edu, an admin creates the app under Workspace settings → Apps → Create. Sign in when prompted.
Grok
Go to grok.com/connectors, choose New Connector → Custom, paste the URL, and complete the sign-in. Available on web, iOS, and Android.
Gemini
Custom connected apps require Gemini Spark (currently US personal accounts, 18+, English, with Keep Activity on): Settings → Connected apps → Add custom app. Everywhere else, the Gemini CLI works today. Add statshawk under mcpServers in settings.json:
{
"mcpServers": {
"statshawk": {
"httpUrl": "https://mcp.statshawk.ai/mcp"
}
}
}Cursor
Open Settings → MCP, click Add new MCP server, and paste:
{
"mcpServers": {
"statshawk": {
"url": "https://mcp.statshawk.ai/mcp"
}
}
}No connector surface? Install the skill.
Agent frameworks and CLI agents (including ones running open models) can use the statshawk agent skill instead: it teaches the agent the statshawk workflows and uses the MCP server when connected or the REST API with your key otherwise:
Claude Code can also install it as a plugin:
/plugin marketplace add EdgeHawk-AI/statshawk-skills, then
/plugin install statshawk@statshawk-skills.