generated: '2026-08-04'
method: probed
source: >-
  Response headers observed on https://lemonperfect.com/api/2026-07/graphql.json
  and the agent policy published at https://lemonperfect.com/agents.md
note: >-
  Lemon Perfect publishes no rate-limit reference page. Everything below was
  observed on a live anonymous request or read from a document the store serves
  itself. No numeric quota is published, so none is asserted.
limits:
- surface: storefront-graphql
  endpoint: https://lemonperfect.com/api/2026-07/graphql.json
  model: query-cost
  limit_count: null
  window: null
  signal_headers: [shopify-complexity-score, shopify-complexity-score-v2]
  signal_body: extensions.cost.requestedQueryCost
  observed:
  - {query: '{shop{name}}', shopify-complexity-score: 0, shopify-complexity-score-v2: 9}
  - {query: '{__schema{queryType{name}}}', requestedQueryCost: 3}
  exhaustion_signal: 'GraphQL THROTTLED error in errors[] (extensions.code)'
  documented: false
  note: >-
    Cost is returned on every response, so an agent can measure its own burn
    rate, but neither the bucket size nor the restore rate is advertised on
    this host.
- surface: ucp-mcp
  endpoint: https://lemonperfect.com/api/ucp/mcp
  model: per-ip
  limit_count: null
  window: null
  exhaustion_signal: 'HTTP 429'
  policy: 'The MCP endpoint is rate-limited per IP. Back off on 429 responses.'
  policy_source: https://lemonperfect.com/agents.md
  documented: true
- surface: storefront-json
  endpoints: ['/products.json', '/collections/{handle}/products.json', '/cart.js', '/search/suggest.json']
  model: unpublished
  limit_count: null
  pagination_cap: 250
  note: 'Read-only JSON endpoints listed in /llms.txt; no rate-limit signal observed in headers.'
x-evidence:
  fetched: '2026-08-04'
  urls:
  - {url: 'https://lemonperfect.com/api/2026-07/graphql.json', http_status: 200}
  - {url: 'https://lemonperfect.com/agents.md', http_status: 200}