generated: '2026-07-21'
method: searched
source: https://docs.ryft.io/integrations/ryft-mcp
status: published
status_raw: published
availability: private-preview
server:
  name: ryft
  description: Ryft MCP is a Model Context Protocol server that enriches AI agents with full lakehouse
    context - table metadata, query history, and domain knowledge - so agents can write accurate queries
    and answer business questions about your Apache Iceberg lakehouse without guessing.
  transport: http
  url: https://app.ryft.io/mcp
  auth:
    type: oauth2
    flow: authorization_code
    authorization_server: https://app.ryft.io/.well-known/oauth-authorization-server
    note: Browser login with Ryft credentials (Auth0). PKCE (S256), dynamic client registration.
clients:
- name: Cursor
  config: '{"mcpServers":{"ryft":{"type":"http","url":"https://app.ryft.io/mcp"}}}'
- name: Claude Code
  install: claude mcp add --transport http ryft https://app.ryft.io/mcp
- name: Claude
  setup: Settings > Connectors > Add custom connector > URL https://app.ryft.io/mcp
tools:
- name: search_tables
  description: Search for Iceberg tables in your lakehouse (documented example tool).
  source: https://docs.ryft.io/integrations/ryft-mcp
notes: Additional tools are exposed by the hosted server but not fully enumerated in public docs; search_tables
  is the one documented by name. Tool list may expand.
deployment:
  mode: remote
  endpoint: https://app.ryft.io/mcp
  verified: probed
  probe: gated
  checked: '2026-08-12'
  source: catalog MCP census