generated: '2026-09-19'
method: searched
source: https://mcp.horizonshield.dev/
docs:
- https://gate.horizonshield.dev/spec
- https://shield.the-horizons-innovation.com/llms.txt
- https://shield.the-horizons-innovation.com/yakumo/plans/
limit_count: 0
summary: >-
  No rate limit is documented for any public API surface, and none was signalled: the KIRA initialize/tools/list,
  gate and ledger responses carried no RateLimit-*, X-RateLimit-* or Retry-After headers, and no 429 was observed.
  The gate spec, the plugin README, the llms.txt and the OpenAPI say nothing about request ceilings. Two adjacent
  facts are recorded so they are not mistaken for API limits: the Yakumo Premium plan describes agent references to a
  member's profile as "within fair use, about 5,000 queries/month as a guide" (a per-member plan allowance, not a
  caller limit), and the gate's watchlist re-measures registered endpoints on a fixed daily cadence (a schedule, not
  a quota). The provider does publish its own traffic — /.well-known/usage-stats.json shows 1,867 external calls in
  the last 30 days — which is a transparency signal rather than a limit.
rate_limits: []
headers_observed:
  kira_mcp_post: [content-type, access-control-allow-origin, access-control-allow-headers, access-control-expose-headers (mcp-session-id), cf-ray, server: cloudflare]
  rate_limit_headers: []
  status_on_exhaustion: unobserved
edge: Cloudflare (all horizonshield.dev hosts and both workers.dev origins)
adjacent_signals:
- {what: 'Yakumo Premium fair-use guide', value: 'about 5,000 agent queries per member profile per month', scope: per-member plan allowance, source: 'https://shield.the-horizons-innovation.com/yakumo/plans/'}
- {what: gate re-measurement cadence, value: daily, scope: per watched endpoint, source: 'https://gate.horizonshield.dev/watchlist'}
- {what: public usage counters, value: '5,438 total external calls; 1,867 in 2026-08-22..2026-09-20', source: 'https://mcp.horizonshield.dev/.well-known/usage-stats.json'}