generated: '2026-09-12'
method: probed
source: >-
  Live probes on 2026-09-12 of https://docs.afterquery.com/mcp,
  https://docs.afterquery.com/.well-known/agent-card.json,
  https://docs.afterquery.com/.well-known/mcp.json, https://www.afterquery.com/llms.txt and the
  /.well-known/ path set across every AfterQuery host.
note: >-
  AfterQuery ships no API contract, so there is no OpenAPI, no securitySchemes and no domain
  standard declared in a contract to assess. What can be asserted is confined to the agent-surface
  standards it actually serves, each with the probe that established it. The AI training-data /
  evaluation market this company sells into has no machine-readable domain standard analogous to
  SCIM, FHIR or ISO 20022, so domain_standard_conformance is not applicable here - and, being
  reward-only, is left unclaimed rather than invented.
conformance:
- id: mcp
  name: Model Context Protocol (JSON-RPC 2.0, streamable HTTP)
  conforms: true
  evidence: >-
    POST https://docs.afterquery.com/mcp with {"jsonrpc":"2.0","id":1,"method":"tools/list"}
    returned HTTP 200 text/event-stream carrying a well-formed JSON-RPC result with three tools,
    each with a JSON Schema inputSchema and MCP tool annotations.
- id: a2a
  name: A2A Agent Card
  conforms: true
  version_declared: '0.3'
  grade: conformant
  evidence: >-
    GET https://docs.afterquery.com/.well-known/agent-card.json returned HTTP 200 application/json
    parsing as an AgentCard object; graded against A2A 1.0.0 hard checks in a2a/afterquery-a2a.yml.
- id: llms-txt
  name: llms.txt
  conforms: true
  evidence: >-
    GET https://www.afterquery.com/llms.txt returned HTTP 200 in llms.txt format (H1, blockquote
    summary, sectioned link lists); a second, separate llms.txt is served for the help centre at
    https://docs.afterquery.com/llms.txt.
- id: agent-skills
  name: Agent Skills (well-known skill.md)
  conforms: true
  evidence: >-
    GET https://docs.afterquery.com/.well-known/agent-skills/afterqueryexperts/skill.md returned
    HTTP 200 with YAML frontmatter (name, description, metadata) over markdown instructions.
- id: openapi
  name: OpenAPI
  conforms: false
  evidence: >-
    No OpenAPI is published. /openapi.json, /openapi.yaml, /swagger.json, /v1/openapi.json,
    /api-docs, /docs and /redoc were probed on www.afterquery.com, docs.afterquery.com and
    api.afterquery.com on 2026-09-12 - every one 404 - and a tree of the documentation MCP's own
    virtual filesystem shows no api-reference directory.
- id: rfc9116-security-txt
  name: RFC 9116 security.txt
  conforms: false
  evidence: /.well-known/security.txt returned 404 on every AfterQuery host probed.
- id: rfc9727-api-catalog
  name: RFC 9727 api-catalog
  conforms: false
  evidence: /.well-known/api-catalog returned 404 on every AfterQuery host probed.
- id: oauth2
  name: OAuth 2.0 / RFC 8414 / RFC 9728
  conforms: false
  evidence: >-
    No authorization-server or protected-resource metadata is served; the one callable surface is
    anonymous (see authentication/afterquery-authentication.yml).
domain_standard:
  applicable: false
  note: >-
    No published domain standard governs the expert-data / model-evaluation market AfterQuery
    sells into. Its benchmarks (IDE-Bench, App-Bench, Market-Bench, FinanceArena, FinanceQA) are
    de-facto evaluation formats it authors itself, not a standard it conforms TO, and its harness
    work runs on Harbor, an upstream framework it forks rather than owns.
certifications:
  published: []
  note: >-
    No SOC 2, ISO 27001, HIPAA, PCI or FedRAMP claim appears on the website, in either llms.txt, or
    on the privacy or terms pages, and there is no trust centre. No Compliance pointer is emitted.