> ## Documentation Index > Fetch the complete documentation index at: https://docs.outzen.ai/llms.txt > Use this file to discover all available pages before exploring further. # API and MCP > Create a personal access token, call the Guhan REST API, and connect an AI client over MCP. You can work with your Guhan workspace from your own code or from an AI client. Both use a **personal access token** that acts as you, with your role and your workspace memberships. ## Create an API key **1. Open your profile** Open your **Profile** and choose the **API keys** tab. API keys belong to you, not to the workspace, so they're not under Settings. ![Profile, API keys: your personal access tokens](https://mintcdn.com/outzen/7ircGJHGowAGODBP/images/api-keys.png?fit=max\&auto=format\&n=7ircGJHGowAGODBP\&q=85\&s=3a3ff36612c1a17149be31a9586ab59f) Profile, API keys: your personal access tokens **2. Create a token** Choose **New token**, give it a name, and pick an expiry or **Never expires**. ![New personal access token: give it a name and an expiry](https://mintcdn.com/outzen/7ircGJHGowAGODBP/images/api-new-token.png?fit=max\&auto=format\&n=7ircGJHGowAGODBP\&q=85\&s=41da7a133405f2f0dfd752dd95e2c11f) New personal access token: give it a name and an expiry **3. Copy it now** The token is shown once. Store it in a password manager or secret store. If you lose it, revoke it and create a new one. ![The token is shown once. Copy it before you dismiss the banner (the value is hidden in this picture)](https://mintcdn.com/outzen/7ircGJHGowAGODBP/images/api-token-banner.png?fit=max\&auto=format\&n=7ircGJHGowAGODBP\&q=85\&s=eed6d976ee7837babf6cbc41755e956a) The token is shown once. Copy it before you dismiss the banner (the value is hidden in this picture) The list shows each token's status (active, expired, or revoked), when it was created, and when it was last used. Revoke a token and anything using it stops working immediately. Tokens start with `pat_live_`. > [!WARNING] > > A token can do everything you can do. It has no narrower scope, so treat it like a password, and use an expiry for anything temporary. ## REST API The public API is documented interactively in the app at `/docs/api`, and the OpenAPI specification is available at `/api/v1/openapi.json`. Send your token as a bearer token: ``` Authorization: Bearer pat_live_... ``` The API covers the main things you do in the app, including campaigns, Lists, prospects, companies, conversations, the Do-Not-Contact list, meetings, products, templates, and brand voice. Every call is checked against the same permissions as the app, so a token can only do what your role allows. ## MCP Guhan runs an MCP server at `https://guhan.ai/api/mcp`. Any MCP-capable AI client — Claude, ChatGPT, Codex, Cursor — can connect to it and work on your workspace through a curated set of tools (check readiness, set up products and templates, build Lists, see where credits went, read the inbox, book meetings, and more). Tools never send outreach, never delete records, and only spend credits against a quote you agreed to. Every call is checked against your role in whichever workspace you're using. Calls are rate limited per token. ### Connect Claude Claude signs in to Guhan with your Guhan account — you don't need to paste a token. **1. Open connector settings** In Claude, open **Settings → Connectors → Add custom connector**. (On Claude Desktop and Claude Web the entry point is the same.) **2. Point it at Guhan** Name it **Guhan** and set the server URL to: ``` https://guhan.ai/api/mcp ``` Leave authentication on the default — Claude discovers Guhan's sign-in automatically. Save and enable the connector. **3. Sign in** The first time you use the connector, Claude opens Guhan's sign-in page in your browser and asks you to approve the connection. Review what the connector can do and choose **Approve**. You're back in Claude once it finishes. **4. Try it** Start a Claude chat with the Guhan connector enabled and ask something like **"What's left to set up in my Guhan workspace?"** Claude calls `check_readiness` and reads your next steps back to you. If you belong to more than one workspace, Claude asks which to use before any tool call. Switching workspaces is a question away — **"Switch to the Acme workspace"** — not a reconnect. ### Connect ChatGPT ChatGPT signs in to Guhan with your Guhan account — like Claude, you don't need to paste a token. **1. Turn on developer mode** In ChatGPT, open **Settings → Connectors** and enable **Developer mode**. (Custom connectors are only visible while Developer mode is on.) **2. Add a custom connector** Choose **Add custom connector** and set the server URL to: ``` https://guhan.ai/api/mcp ``` Name it **Guhan**. Save the connector. **3. Sign in** The first time you use the connector, ChatGPT opens Guhan's sign-in page in your browser and asks you to approve the connection. Review what the connector can do and choose **Approve**. **4. Try it** Start a ChatGPT chat with the Guhan connector enabled and ask something like **"What's left to set up in my Guhan workspace?"** ChatGPT calls `check_readiness` and reads your next steps back to you. If you belong to more than one workspace, ChatGPT asks which to use before any tool call. Switching workspaces is a question away — **"Switch to the Acme workspace"** — not a reconnect. ### Connect other clients * **Codex** — run `codex mcp add guhan https://guhan.ai/api/mcp`, then `codex mcp login guhan`. * **Clients that can't do OAuth** — create a Personal Access Token above and paste it as a bearer token. ## If something's off * **401 Unauthorized.** The token is missing, mistyped, expired, or revoked. Create a new one. * **403 Forbidden.** Your role doesn't allow that action. See [Roles and permissions](https://docs.outzen.ai/workspace-settings/roles-and-permissions.md). * **Rate limited.** MCP calls are limited per token; slow down and retry. ## Related > [card] [Roles and permissions](https://docs.outzen.ai/workspace-settings/roles-and-permissions.md) > > What a token can do depends on your role. > [card] [Ask Guhan](https://docs.outzen.ai/ask-guhan.md) > > Prefer chat? Ask in plain language. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.