Clarisights · Authentication Profile

Clarisights Authentication

Authentication

Clarisights secures its APIs with delegated-user-login across 1 declared security scheme, as derived from its OpenAPI definitions.

CompanyMarketing AnalyticsMarketing IntelligenceBusiness IntelligenceData IntegrationReportingDashboardsMCPAI
Methods: delegated-user-login Schemes: 1 OAuth flows: API key in:

Security Schemes

clarisights-workspace-login delegated-user-login

Source

Authentication Profile

clarisights-authentication.yml Raw ↑
generated: '2026-08-13'
method: searched
source: https://clarisights.com/marketing-data-mcp
sources:
- https://clarisights.com/marketing-data-mcp
- https://clarisights.com/llms.txt
- https://github.com/clarisights/claude-plugins
note: >-
  Derived-from-spec was not possible: Clarisights publishes no OpenAPI, so
  derive-authentication.py produced zero schemes (0 spec files). This profile is
  SEARCHED from the provider's own Marketing Data MCP documentation, which
  describes the auth model for the only programmatic surface the company ships.
summary:
  types: [delegated-user-login]
  api_key_in: []
  oauth2_flows: []
  api_keys_issued: false
  api_keys_note: >-
    Explicitly none. The docs say it twice — "no separate tokens to manage" and
    "No separate tokens, no parallel permission tree." There is no developer
    key-issuance surface anywhere on the site.
schemes:
- name: clarisights-workspace-login
  type: delegated-user-login
  surface: mcp
  sources: [https://clarisights.com/marketing-data-mcp]
  description: >-
    The MCP client (Claude, ChatGPT, Cursor, Gemini, or any MCP-compatible tool)
    is pointed at a workspace-scoped connection URL and the end user signs in
    with their own Clarisights credentials from inside the AI client. Every call
    is then made on that logged-in user's behalf, using their auth, their
    permissions, and their history.
  authorization_model: acl-inherited
  authorization_detail: >-
    "Whatever that user can see in Clarisights, the AI can see: accounts,
    regions, brands, reports. Nothing more." There is no parallel scope or
    permission tree for the agent surface — the workspace ACL is the authority.
  admin_gate:
    control: Settings -> Workspace -> AI access
    granularity: global on/off for the whole workspace
    actor: workspace admin
  enrollment:
  - step: 1
    action: Workspace admin enables AI access (Settings -> Workspace -> AI access).
  - step: 2
    action: Signed-in user opens their MCP settings page and copies the workspace connection URL.
  - step: 3
    action: User pastes the URL into their AI client and signs in with their Clarisights account.
  scopes: []
  scopes_note: >-
    No OAuth scope surface is published, so scopes/ is deliberately not emitted.
    Authorization is entirely ACL-inherited rather than scope-granted.
sso:
  supported: true
  evidence: >-
    The Effective Self-Service page (linked from the provider llms.txt) states
    "row/column-level security with SSO". No SSO protocol, IdP list, or metadata
    endpoint is published publicly.
  protocols_published: false
oidc_discovery:
  probed: true
  result: none
  detail: >-
    /.well-known/openid-configuration, /.well-known/oauth-authorization-server
    and /.well-known/oauth-protected-resource all return 404 on clarisights.com
    and 404 on app.clarisights.com (probed 2026-08-13). See
    well-known/clarisights-well-known.yml.
data_handling:
  stores_user_data: false
  evidence: >-
    "The MCP server is an interface, not a store. Your data stays in
    Clarisights." Documented GDPR-compliant with no region-specific data
    residency wrinkles.