> ## Documentation Index
> Fetch the complete documentation index at: https://docs.reply.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Reply MCP

> Connect Reply to your AI client and let it work across your team/workspace.

Reply MCP exposes your Reply team/workspace to MCP-compatible AI clients — Claude, ChatGPT, Cursor, Codex, and more. One connection gives an agent a catalog of tools spanning your CRM and your AI SDR, so it can read and act on Reply data without making direct HTTP requests. The tools run the same logic as the Reply API and the in-app AI assistant.

## What you can do

Once connected, an agent can work across these areas of Reply.

<CardGroup cols={2}>
  <Card title="Contacts & lists" icon="address-book">
    Find, create, update, and organize contacts and lists.
  </Card>

  <Card title="Sequences" icon="list-check">
    Inspect sequences and steps; enroll contacts; start or pause; assign channels and schedules.
  </Card>

  <Card title="Analytics" icon="chart-line">
    Measure sequence performance across email and LinkedIn — for one sequence or compared side by side, over any date range.
  </Card>

  <Card title="Inbox" icon="inbox">
    Read inbox emails, send replies, and re-categorize threads.
  </Card>

  <Card title="Tasks" icon="circle-check">
    List, create, and complete tasks.
  </Card>

  <Card title="Channels & accounts" icon="envelope">
    List connected email and LinkedIn accounts.
  </Card>

  <Card title="Team" icon="user-group">
    Look up team members.
  </Card>

  <Card title="Knowledge & help" icon="book-open">
    Search the knowledge base, read articles, and explore the app.
  </Card>
</CardGroup>

### AI SDR

The largest set of tools drives Reply's AI SDR — the autonomous agent that researches, writes, and replies on your behalf. Through MCP, an agent can build and manage everything the AI SDR runs on:

<CardGroup cols={2}>
  <Card title="Offers" icon="tag">
    Create and refine the offers your AI SDR pitches.
  </Card>

  <Card title="Playbooks" icon="book">
    Author, update, and reuse the playbooks that guide outreach.
  </Card>

  <Card title="Knowledge bases" icon="database">
    Build and maintain the knowledge the AI SDR draws on.
  </Card>

  <Card title="Reply handlers" icon="reply">
    Configure how the AI SDR responds to incoming replies.
  </Card>

  <Card title="Re-engagement" icon="arrows-rotate">
    Set up cards that bring cold contacts back into play.
  </Card>

  <Card title="Approvals & reply mode" icon="thumbs-up">
    Review, approve, or regenerate AI-drafted messages and control reply mode.
  </Card>
</CardGroup>

## Authentication

Reply MCP authenticates as a single Reply user. Connect with one of two methods.

### Personal API key (recommended)

Authenticate with your personal API key, passed as `Authorization: Bearer <token>`. This is the preferred method because a personal key is **scoped** — you grant it only the permissions your agent needs, and it can call only the tools those permissions cover.

A scope is a permission such as `contacts:read` or `sequences:operate`. Each tool requires one scope, and a key can call a tool only if it holds that scope. Scopes are grouped by area — contacts, sequences, inbox, tasks, channels, and AI SDR — mirroring the [capabilities](#what-you-can-do) above, so you can grant an agent exactly the reach it needs and nothing more. See [Scopes](/api-reference/authentication#scopes) for the full list.

Create and scope a key under **Settings → API Key** in your [Reply.io dashboard](https://run.reply.io). Keep the token private.

### OAuth

If your client supports OAuth, authorize the connection directly from the client — no key to copy or store. OAuth grants access to the **full tool catalog** (all scopes), so prefer a scoped personal API key when you want to limit what the agent can do.

<Note>
  Some OAuth clients don't support Client ID Metadata Documents (CIMD) and fall back to Dynamic Client Registration, which is allow-listed on our side. If your client reports an allow-list or registration error when connecting, let us know in the [community Slack](/slack) and we'll review it.
</Note>

### Unsupported keys

Reply MCP is a single-user connection, so **act-on-behalf API keys are not supported** — including team-level **Team keys** and **Organization API keys**. These are designed to make requests on behalf of other users, which a single-user connection doesn't provide. Use a personal API key or OAuth instead.

<Note>
  Passing credentials in the URL (for example `?api_key=...`) is not supported. Always authenticate with the `Authorization` header or OAuth.
</Note>

## Connect

Reply MCP is a remote MCP server at `https://mcp.reply.io`. Once you've chosen an [authentication](#authentication) method, add it to your client.

<Steps>
  <Step title="Add Reply MCP to your client">
    Pass your personal API key in the `Authorization` header:

    <CodeGroup>
      ```bash Claude Code theme={null}
      claude mcp add --transport http reply-mcp https://mcp.reply.io \
        --header "Authorization: Bearer usk_...your_key..."
      ```

      ```bash Codex theme={null}
      export REPLY_MCP_TOKEN="usk_...your_key..."
      codex mcp add reply-mcp --url https://mcp.reply.io \
        --bearer-token-env-var REPLY_MCP_TOKEN
      ```
    </CodeGroup>

    **Desktop apps (Claude, ChatGPT):** add Reply MCP as a remote/custom connector using the URL `https://mcp.reply.io` and authorize with OAuth.
  </Step>

  <Step title="Verify">
    Ask your client to list its available tools. You should see the `reply_*` tools, ready to use.
  </Step>
</Steps>

<Accordion title="Legacy: connect through the mcp-remote proxy">
  If your client only supports local (stdio) MCP servers, bridge to the remote server with the `mcp-remote` proxy. Prefer a native remote connection where possible.

  ```json theme={null}
  {
    "mcpServers": {
      "reply": {
        "command": "npx",
        "args": [
          "-y",
          "mcp-remote",
          "https://mcp.reply.io",
          "--header",
          "Authorization: Bearer usk_...your_key..."
        ]
      }
    }
  }
  ```
</Accordion>

## Security

* Treat your API token like a password.
* Do not share it in public repositories, screenshots, or chat messages.
* Scope tokens to the minimum your agent needs.
* If you think a token has been exposed, revoke it in Reply and create a new one.

## Related

* [Authentication](/api-reference/authentication) — generate and manage API keys.
* [Connecting Reply MCP to your AI model (Help Center)](https://support.reply.io/en/articles/12955018-connecting-reply-mcp-to-your-ai-model)
* [Reply MCP overview](https://reply.io/mcp/)