generated: '2026-09-19'
method: probed
status: published
source: https://avalix.ai/autonoma/mcp
docs: https://avalix.ai/autonoma/agents
summary: >-
  Avalix operates ONE remote MCP server at https://avalix.ai/autonoma/mcp — the same origin as its website,
  the Autonoma OpenAPI and the A2A JSON-RPC endpoint. It is POST-only Streamable HTTP (a GET returns the
  JSON 404 {"error": "not found"}), reports MCP protocol version 2025-11-25, identifies itself as serverInfo
  {name: ai.avalix/autonoma, title: Avalix Autonoma Agent Services, version: 1.0.0, websiteUrl:
  https://avalix.ai/autonoma/}, and answers initialize and tools/list anonymously with six tools carrying
  real inputSchemas. resources/list and prompts/list return -32601 Method not found. The server's own
  initialize instructions read: "Avalix-owned service endpoint. Discover fixed-price services, run free
  previews, and request paid work. Never provide secrets." The tool names and descriptions are
  Autonoma-specific (list_offers, trust_preview, quote_task with an authorization_attestation const,
  request_service keyed on offer_slug) and match no shared platform fingerprint — a provider-built server.
  No MCP host of its own exists (mcp.avalix.ai does not resolve); the MCP resource host is the apex.
deployment:
  mode: remote
  endpoint: https://avalix.ai/autonoma/mcp
  auth: none
  verified: probed
  note: >-
    A hosted HTTPS endpoint an MCP client POSTs to directly; Avalix ships no stdio package and documents no
    npx/uvx install (npm and PyPI searched 2026-09-19, nothing first-party). auth is "none" in the connection
    sense — initialize and tools/list need no credential, and no OAuth metadata is served for this resource
    (/.well-known/oauth-authorization-server and /.well-known/oauth-protected-resource return 404 on the apex
    and JSON 404 under /autonoma/). Paid work is gated by x402 PAYMENT: request_service without a
    payment_reference "returns exact Base USDC payment requirements; retry with the confirmed transaction
    hash". That is an economic gate on invocation, not an authentication scheme on the server.
servers:
- id: autonoma
  name: Avalix Autonoma Agent Services
  endpoint: https://avalix.ai/autonoma/mcp
  transport: streamable-http
  http_methods: [POST]
  auth: none
  payment: x402 (USDC on Base, eip155:8453) per paid service; request_service returns payment requirements, retried with a transaction hash
  status: live
  probe:
    fetched: '2026-09-19'
    initialize:
      http_status: 200
      protocol_version: '2025-11-25'
      server_info: {name: ai.avalix/autonoma, title: Avalix Autonoma Agent Services, version: 1.0.0, websiteUrl: 'https://avalix.ai/autonoma/'}
      capabilities: {tools: {listChanged: false}}
      instructions: 'Avalix-owned service endpoint. Discover fixed-price services, run free previews, and request paid work. Never provide secrets.'
    tools_list:
      http_status: 200
      content_type: application/json
      tool_count: 6
      file: mcp/avalix-ai-mcp-tools.json
    resources_list: {http_status: 200, error: '-32601 Method not found'}
    prompts_list: {http_status: 200, error: '-32601 Method not found'}
    get_request: {http_status: 404, body: '{"error": "not found"}'}
tools:
- name: compatibility_check
  title: Free MCP, A2A, or OpenAPI compatibility check
  description: Run deterministic static checks against one public URL or supplied non-sensitive JSON document.
  input_schema: '{kind: enum[auto,a2a,mcp,openapi], url: string(uri), document: object}; additionalProperties false'
  cost: free
  source_operation: null
  note: No REST twin in the OpenAPI; the same free check is offered on the web at https://avalix.ai/autonoma/check and as A2A skill compatibility-check.
- name: list_offers
  title: List paid agent services
  description: Return current fixed-price services and verified performance totals.
  input_schema: '{} (no properties)'
  cost: free
  source_operation: 'openapi/avalix-ai-openapi.json#GET /v1/offers (no operationId)'
- name: trust_preview
  title: Free agent trust preview
  description: Run a free deterministic preview of supplied agent evidence.
  input_schema: 'object, additionalProperties true (the REST twin requires agent_card to be a JSON object)'
  cost: free
  source_operation: 'openapi/avalix-ai-openapi.json#POST /v1/trust/preview (no operationId)'
- name: task_capabilities
  title: Discover task capabilities
  description: List bounded digital work this agent can quote and deliver.
  input_schema: '{} (no properties)'
  cost: free
  source_operation: 'openapi/avalix-ai-openapi.json#GET /v1/tasks/capabilities (no operationId)'
- name: quote_task
  title: Quote a custom task
  description: Classify and price a bounded task. Requires authority to commission the work and a target scope.
  input_schema: 'required [request(maxLength 12000), target_scope(maxLength 2000), authorization_attestation(const true)]; deliverables: array(maxItems 10)'
  cost: free (quoting); the quoted work is paid
  source_operation: 'openapi/avalix-ai-openapi.json#POST /v1/tasks/quote (no operationId)'
- name: request_service
  title: Request paid agent work
  description: Request a service. Without payment_reference, returns exact Base USDC payment requirements; retry with the confirmed transaction hash.
  input_schema: 'required [offer_slug]; payment_reference, payer_address, target_scope: string; authorization_attestation: boolean; additionalProperties true'
  cost: per offer_slug (see plans/avalix-ai-plans-pricing.yml)
  source_operation: 'openapi/avalix-ai-openapi.json#trust_preflight | agent_card_lint | receipt_verifier | data_distill | agent_risk_scan | agent_audit | agent_security_test | research_brief | monitor | integration_warranty_7d | mcp_compatibility | openapi_repair | agent_contract — one paid route per offer_slug'