MCP

Connect an AI assistant

Pick where your Stirling PDF runs, then add its endpoint to your client.

Endpoint: https://api.stirling.com/mcp

Open Settings → Developer → MCP Server. It shows Your MCP endpoint and, under Connect your AI assistant, a snippet for each client. Paste it into the file shown, then restart the client. On first use you sign in with your Stirling account and click Allow access on the Authorize access page. There is no key to copy.

Claude Code (.mcp.json). For VS Code, use the same in .vscode/mcp.json with servers in place of mcpServers:

json

{"mcpServers": {"stirling-pdf": {"type": "http", "url": "https://api.stirling.com/mcp"}}}

Codex CLI (~/.codex/config.toml):

toml

[mcp_servers.stirling-pdf]
url = "https://api.stirling.com/mcp"

Claude Desktop (claude_desktop_config.json) reaches remote servers through the mcp-remote bridge, which needs Node.js:

json

{
  "mcpServers": {
    "stirling-pdf": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://api.stirling.com/mcp"]
    }
  }
}

Guests can't connect MCP clients. Every tool call counts towards your team's usage, like API calls (Plans, credits and billing).

Endpoint: https://your-server/mcp, on the same host and port as Stirling PDF.

MCP is off by default. An admin turns it on and picks how clients sign in (MCP server settings). OAuth is the default, so ask your admin which mode is set before you use an API key.

API key mode: create a key under Settings → Preferences → API Keys and send it as X-API-KEY: <key> or Authorization: Bearer <key>.
OAuth mode: add the endpoint with no key. The client finds your identity provider at /.well-known/oauth-protected-resource and signs you in.

Claude Code in API key mode (.mcp.json):

json

{
  "mcpServers": {
    "stirling-pdf": {
      "type": "http",
      "url": "https://your-server/mcp",
      "headers": { "X-API-KEY": "your-stirling-api-key" }
    }
  }
}

Claude Desktop through mcp-remote (needs Node.js). In OAuth mode, leave out the --header line and mcp-remote runs the sign-in:

json

{
  "mcpServers": {
    "stirling-pdf": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote", "https://your-server/mcp",
        "--header", "X-API-KEY:your-stirling-api-key"
      ]
    }
  }
}

To test the connection, run npx @modelcontextprotocol/inspector, choose the streamable HTTP transport, enter your /mcp URL and add the same header.

PDF tool calls count as automation, so they use the server's allowance even with a Team licence (Processing allowance).

Edit this page on GitHub