generated: '2026-08-12'
method: probed
source: >-
  Response headers observed on live unauthenticated requests to
  https://mcp.valid.co/health, https://mcp.valid.co/api/mcp/ and
  https://api.valid.co/ on 2026-08-12, plus a search of valid.co for any published
  limits.
limit_count: 0
limits: []
headers: []
status_on_exhaustion: null
notes: >-
  Valid documents no rate limits anywhere on its public surface — there is no
  developer portal, no API reference and no mention of limits in the terms of
  service. Nor are limits signalled at runtime: no RateLimit-*, X-RateLimit-*, or
  Retry-After header appeared on any response from mcp.valid.co (repeated requests
  to /health), from the MCP endpoint's 401, or from the Kong gateway at
  api.valid.co. An agent calling this surface has no way to learn its budget in
  advance and no runtime signal to back off on — it would have to infer throttling
  from failures. Recording the honest zero: nothing published, nothing signalled.
  This says nothing about whether limits are enforced behind the auth gate, only
  that they are not observable.