API Evangelist enhancements — CharacterQuilt Discovery API

# generated: '2026-08-13'
# method: generated
# source: openapi/characterquilt-discovery-api-openapi.yml + live probes 2026-08-13
# note: >-
#   OpenAPI Overlay 1.0.0 capturing API Evangelist enhancements to the brand-profiles
#   Discovery API description. Applied on top of the original; the original is never
#   mutated. Every value added here was observed on the wire.
overlay: 1.0.0
info:
  title: API Evangelist enhancements — CharacterQuilt Discovery API
  version: 1.0.0
extends: openapi/characterquilt-discovery-api-openapi.yml
actions:
- target: $.info
  description: Record provenance and the discovery posture.
  update:
    x-provenance:
      authored-by: API Evangelist
      method: generated
      basis: https://www.characterquilt.com/llms.txt observed HTTP 200, 350254 bytes,
        6251 lines
      note: >-
        /llms.txt is CharacterQuilt's only published machine-readable discovery
        document. Every /.well-known path on this host returns 404; the single
        /.well-known hit anywhere on the company's infrastructure is the RFC 9728
        OAuth protected-resource metadata on mcp.characterquilt.com, which describes
        a different surface entirely.
    x-artifacts:
      llms-txt: llms/characterquilt-llms.txt
      well-known: well-known/characterquilt-well-known.yml
      conventions: conventions/characterquilt-conventions.yml
- target: $
  description: State the security posture explicitly.
  update:
    security: []
    x-authentication: none
- target: $.paths['/llms.txt'].get
  description: Record the real size and the absence of pagination — this is the whole
    catalog in one response.
  update:
    x-payload:
      bytes: 350254
      lines: 6251
      observed: '2026-08-13'
      note: >-
        The entire catalog is delivered as a single document. There is no paginated
        listing, no index endpoint, no cursor and no Link header — a consumer that
        wants one slug must either already know it or download 350 KB.
    x-caching:
      note: Use conditional requests; the document is served with standard HTTP caching
        headers.
    x-rate-limit:
      documented: false
      headers: []
- target: $.paths['/llms.txt'].get.responses['200']
  description: Document the llms.txt structure so an agent knows how to parse it.
  update:
    x-format:
      standard: llms.txt
      structure: >-
        H1 title, blockquote summary, "## How to use" section describing the
        /branding/<slug> and /branding/<slug>.json convention, then "## Brand
        profiles" — a flat markdown link list of one entry per company in the form
        "- [Name](/branding/slug): #PRIMARYHEX · Font Family".
      derivable: >-
        Each line yields a company name, a slug (and therefore a profile URL), a
        primary brand colour and a primary font without fetching the profile.