generated: '2026-08-13'
method: probed
status: published
source: https://sessionai.com/mcp
server:
  name: Session AI
  transport: http
  url: https://sessionai.com/mcp
deployment:
  mode: remote
  endpoint: https://sessionai.com/mcp
  auth: oauth
  verified: probed
evidence:
  fetched: '2026-08-13'
  get:
    url: https://sessionai.com/mcp
    http_status: 401
    content_type: application/json
    body: '{"error":"unauthorized"}'
    www_authenticate: >-
      Bearer realm="mcp", resource_metadata="https://sessionai.com/.well-known/oauth-protected-resource"
    access_control_expose_headers: WWW-Authenticate, Mcp-Session-Id, Mcp-Protocol-Version
  post_tools_list:
    url: https://sessionai.com/mcp
    method: POST
    body: '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
    http_status: 401
  control_probe:
    url: https://sessionai.com/zzz-not-a-real-path
    http_status: 404
    note: >-
      Control path returns 404, so /mcp is a genuine dedicated route and not a
      single-page-app catch-all answering 200/401 to everything.
authorization:
  protected_resource_metadata: https://sessionai.com/.well-known/oauth-protected-resource
  resource: https://sessionai.com/mcp
  resource_name: Session AI
  authorization_servers:
  - https://jsaqyxhgmkgcyaocjnzz.supabase.co/auth/v1
  bearer_methods_supported:
  - header
  dynamic_client_registration: https://jsaqyxhgmkgcyaocjnzz.supabase.co/auth/v1/oauth/clients/register
  spec: RFC 9728 OAuth 2.0 Protected Resource Metadata
tools: []
tools_gated: true
tools_note: >-
  tools/list returns HTTP 401 to an anonymous caller, so the live tool set and per-tool
  inputSchema could not be enumerated. Session AI publishes no MCP tool list in its docs
  or llms.txt either, so nothing is recorded here rather than guessing. Enumerating the
  tools requires an authenticated bearer token obtained through the Supabase-hosted
  authorization server named in the protected-resource metadata.
ownership_note: >-
  The endpoint is served from Session AI's own apex domain and its protected-resource
  metadata self-identifies as "Session AI", so the server belongs to this company.