generated: '2026-07-21' method: searched status: published source: https://www.supper.co/mcp server: name: supper description: Supper exposes a hosted, remote MCP server so any MCP-compatible AI assistant (Claude, Claude Code, etc.) can query a company's connected data in natural language, grounded in Supper's semantic model. transport: http url: https://api.supper.co/mcp auth: type: oauth2 grant_types: - authorization_code - refresh_token pkce: S256 dynamic_client_registration: true authorization_endpoint: https://api.supper.co/mcp/authorize token_endpoint: https://api.supper.co/mcp/token registration_endpoint: https://api.supper.co/mcp/register metadata: https://api.supper.co/.well-known/oauth-authorization-server notes: One-click OAuth for Claude connections (no API keys or config files). For backend applications the OAuth access token can be read from the macOS keychain (stored by Claude Code) or passed as an environment variable. setup: 'In Claude: Customize -> Connectors -> find Supper in the connectors list and authorize via OAuth (~two-minute setup).' tools: - name: ask description: Ask any data question in natural language. Returns a verified answer grounded in your semantic model. - name: get_pipeline_status description: Check the processing status for complex / long-running queries. - name: get_answer_rows description: Retrieve structured row data from a completed answer - columnar format, ready to act on. - name: get_context description: Retrieve semantic model context - metric definitions, entity descriptions, available tables. notes: Tool names and descriptions are quoted from Supper's public /mcp page. The live MCP endpoint (https://api.supper.co/mcp) and its OAuth metadata (RFC 8414) were verified by probe; tool input schemas are not published and are not fabricated here. deployment: mode: remote endpoint: https://api.supper.co/mcp verified: probed probe: gated checked: '2026-08-12' source: catalog MCP census