generated: '2026-09-09'
method: probed
source: >-
  https://transcriptfetch.com/.well-known/oauth-protected-resource and
  https://clerk.transcriptfetch.com/.well-known/oauth-authorization-server (both HTTP
  200, fetched 2026-09-09). The REST OpenAPI declares only bearerAuth (API key), so
  derive-oauth-scopes.py found no oauth2 scheme; these scopes come from the served
  OAuth metadata, which governs the MCP surface.
docs: https://transcriptfetch.com/docs/mcp
surface: MCP server (https://transcriptfetch.com/mcp) via Clerk-run OAuth; REST API uses bearer API keys without scopes.
scopes:
- name: openid
  description: OpenID Connect authentication (standard OIDC scope).
- name: profile
  description: Access to basic profile claims (name, picture, preferred_username).
- name: email
  description: Access to the account email address and verification status.
- name: offline_access
  description: Refresh-token issuance for long-lived MCP client connections.
- name: public_metadata
  description: Clerk user public metadata (authorization-server metadata; not documented per-endpoint by TranscriptFetch).
- name: private_metadata
  description: Clerk user private metadata (authorization-server metadata; not documented per-endpoint by TranscriptFetch).
notes:
- The protected-resource document scopes_supported lists openid, profile, email, offline_access for the MCP resource.
- Identity scopes only — API authorization is account/credit-based, not scope-partitioned; there are no per-tool or per-endpoint scopes.