generated: '2026-08-13'
method: probed
source: https://mcp.mutinyhq.com/.well-known/oauth-authorization-server
docs: https://help.mutinyhq.com/articles/5003451538-connecting-mutiny-to-claude
api: Mutiny MCP Server
authorization_server: https://mcp.mutinyhq.com
protected_resource: https://mcp.mutinyhq.com
flows:
- type: authorization_code
  authorization_endpoint: https://mcp.mutinyhq.com/oauth/authorize
  token_endpoint: https://mcp.mutinyhq.com/oauth/token
  pkce: S256
scope_count: 5
scopes:
- name: read_asset_groups
  description: >-
    Read the caller's Mutiny asset groups — the deal rooms, landing pages, decks, proposals and other
    customer-facing assets in the workspace. Mutiny's own connector consent screen renders this as
    "Read assets".
  consent_label: Read assets
  access: read
- name: create_asset_groups
  description: >-
    Create new Mutiny asset groups and iterate on them (generate a page/deck/proposal from a template
    plus conversation context, then revise sections or headlines). Consent screen: "Create assets".
  consent_label: Create assets
  access: write
- name: publish_asset_groups
  description: >-
    Publish an asset group to production and return a live, shareable URL. This is the scope that makes
    an agent's output externally visible to a prospect, so it is the highest-consequence scope in the
    set. Consent screen: "Publish assets".
  consent_label: Publish assets
  access: write
- name: read_library_content
  description: >-
    Search and list the workspace content library — logos, images, case studies and other reusable brand
    assets, including filtering by tag. Consent screen: "Read library content".
  consent_label: Read library content
  access: read
- name: manage_library_content
  description: >-
    Upload items into the content library from a URL and manage their organisation, including tagging
    library items. Consent screen: "Manage library content".
  consent_label: Manage library content
  access: write
notes: >-
  The five scope strings are read verbatim from Mutiny's RFC 8414 authorization-server metadata and are
  repeated identically in its RFC 9728 protected-resource metadata (both saved under well-known/). The
  human-readable consent labels are quoted from the Mutiny help-centre article on connecting Mutiny to
  Claude. Mutiny publishes no separate scopes/permissions reference page, so the descriptions above
  restate what each scope name plus the documented MCP capabilities cover; no scope is invented, and no
  scope beyond these five is claimed. There is no default/implicit scope and no refresh_token grant
  advertised — the metadata declares grant_types_supported: [authorization_code] only.