Polar MCP uses OAuth. When you connect a client, your browser opens so you can sign in and authorize access to your organization. You never need to copy an API key into your agent.
Connect your agent
Use the production server athttps://mcp.polar.sh/mcp/polar-mcp for your live organization. While building or testing, use the sandbox server at https://mcp.polar.sh/mcp/polar-sandbox and give it a distinct name such as polar-sandbox.
Choose your client and add the appropriate server.
- Cursor
- Claude Code
- ChatGPT
- Claude Desktop
- Codex
- OpenCode
- Other clients
Add the following to Open Cursor’s MCP settings and connect to
.cursor/mcp.json in your project, or to ~/.cursor/mcp.json to make Polar available in every project:polar to complete authorization.How dynamic toolsets work
Polar supports a broad set of operations. Sending every operation and its full input schema to an agent at the start of every conversation would consume context, make tool selection less reliable, and add unnecessary latency. Instead, Polar uses a dynamic toolset with three lightweight meta-tools. Your agent discovers the relevant operation at runtime, loads only the schema it needs, and then executes it.1
Search
The agent calls
search_tools with what it wants to accomplish, such as “find a customer and check their active subscriptions.”2
Inspect
The agent calls
describe_tools for the best matches to learn their exact inputs and behavior.3
Execute
The agent calls
execute_tool with the selected tool name and validated arguments.Available tools
The runtime tools above provide access to the following Polar operations.Available tools (100)
Available tools (100)

