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

# Embedded Servers

> Built-in platform MCP servers — no connections required; authenticate and call

**Type:** Embedded | **PCID required:** No

Embedded servers are built into the Pinkfish platform. They need no external connections or PinkConnect setup—authenticate with your API key and call them like any other MCP endpoint.

They cover platform capabilities such as **agents and workflows**, **code and browser automation**, **structured and unstructured storage**, **search and web utilities**, **hosted AI model endpoints**, **document and media processing**, and **small deployment utilities** (charts, simple web apps, email, and similar).

**Built-in agent tools** (e.g. creating and downloading artifacts) ship with the agent runtime itself—they are not separate MCP servers.

## Calling pattern

Every embedded server uses the same JSON-RPC 2.0 pattern. Replace `<server-path>` with the route for the server you want (see the sidebar or call `tools/list` on that path):

```bash theme={null}
curl -s -X POST "https://mcp.pinkfish.ai/<server-path>" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "<tool_name>",
      "arguments": { ... }
    }
  }'
```

For authentication, see the [Authentication guide](/api-reference/platform/authentication).

## Discovering servers and tools

* **Sidebar** — Under **Embedded MCP Servers**, each entry is a dedicated page with paths, tools, and parameters.
* **At runtime** — `POST` to a server path with `"method": "tools/list"` and empty `params` to list tools and schemas for that server.

For a broader picture of how MCP fits into Pinkfish, see [Connections](/api-reference/platform/connections) and [MCPs in agents](/agents/mcps).