generated: '2026-08-28' method: probed source: >- https://www.group1auto.com/robots.txt (200) and response headers observed on https://www.group1auto.com/llms.txt (200), 2026-08-28. limit_count: 1 note: >- Group 1 Automotive publishes no API rate limit, because it publishes no API program. The one throughput constraint it does state for automated clients is a robots.txt Crawl-delay, which applies to the same host that serves the agent-facing /llm/inventory/ endpoint. It is recorded here as what it is — a crawl directive, not an HTTP rate-limit contract — so a reader is not left thinking the surface is unbounded, and not left thinking it is metered either. limits: - scope: per-client, host-wide (all automated user agents) surface: https://www.group1auto.com/ (including /llm/inventory/) kind: robots.txt crawl-delay window: 1 second limit: 1 request burst: null enforced_by: >- Voluntary compliance. Not enforced with a 429 — the observable enforcement on this host is Cloudflare bot management, which answers 403 to non-browser clients regardless of pacing. source: https://www.group1auto.com/robots.txt directive: 'Crawl-delay: 1' response_headers: ratelimit_standard: [] x_ratelimit: [] retry_after: false observed_on: - url: https://www.group1auto.com/llms.txt status: 200 headers_present: - cache-control - strict-transport-security - x-envoy-upstream-service-time - cf-cache-status rate_limit_headers_present: false exhaustion: status_code: null detail: >- No 429 was observed and none is documented. The refusal an over-eager or non-browser client actually receives on this host is HTTP 403 with a Cloudflare interstitial, which is a bot decision rather than a rate decision and carries no Retry-After. gaps: - No documented requests-per-second, per-minute or per-day figure for /llm/inventory/. - No RateLimit-* or X-RateLimit-* headers, so an agent cannot pace itself from the response. - No 429 semantics and no Retry-After, so a blocked agent has no documented backoff.