generated: '2026-09-19'
method: probed
source: >-
  Response headers observed on https://www.elderlycarematch.com/a2a/v1 (GET descriptor, POST SendMessage,
  POST tasks/get) and https://www.elderlycarematch.com/.well-known/agent-card.json on 2026-09-19.
docs:
- https://www.elderlycarematch.com/llms.txt
limit_count: 0
summary: >-
  No rate limit is published and no rate-limit header is returned. llms.txt and the agent card document the
  per-call cap (limit <= 20 results per find-care-facilities call) but no request ceiling, and none of the
  seven standard headers appeared on any observed response. An honest zero measured from live responses.
limits: []
per_call_caps:
- {skill: find-care-facilities, parameter: limit, max: 20, source: agent card skill description}
response_headers:
  observed: []
  checked_for: [RateLimit-Limit, RateLimit-Remaining, RateLimit-Reset, X-RateLimit-Limit, X-RateLimit-Remaining, X-RateLimit-Reset, Retry-After]
  note: >-
    None appeared. Responses carry only edge headers (server Vercel, x-vercel-id, x-vercel-cache,
    x-railway-edge, x-railway-request-id, x-hikari-trace) and Cache-Control. An agent gets no pacing signal
    and would learn a limit only by being refused.
exhaustion_status: null
enforcement:
  documented: false
  inferred: >-
    The origin is an Express app on Railway behind Vercel's edge (per response headers), both of which can
    throttle; nothing states that they do. Recorded as inference, not as a published limit.
x-evidence:
  fetched: '2026-09-19'
  probes:
  - {url: 'https://www.elderlycarematch.com/a2a/v1', method: POST, status: 200, rate_limit_headers: none}
  - {url: 'https://www.elderlycarematch.com/a2a/v1', method: GET, status: 200, rate_limit_headers: none}
  - {url: 'https://www.elderlycarematch.com/.well-known/agent-card.json', status: 200, rate_limit_headers: none, cache_control: 'public, max-age=300'}