# AgenticWebScore — AI access declaration # Companion to robots.txt, which carries the same Content-Signal in every group. site: https://agenticwebscore.com contact: hello@agenticwebscore.com policy: https://agenticwebscore.com/ai-policy crawl: allowed search-index: allowed ai-input: allowed ai-train: not allowed attribution: required rate-limit: none on documents, and none on pages except /verify/ crawl-delay: none declared # Documents carry no rate limit, and neither do pages except /verify/. # robots.txt carries no Crawl-delay line. The policy is to fetch what you need at # a reasonable rate. # These are limited in the memory of the process that serves them, and # the counters reset when it restarts: # Scans: 10 per client per minute and 3 running at once, across every entry point that starts one. A reused record or a joined scan spends nothing. # /api/health: 30 per client per minute. # Badge claims: 10 per client per minute. # Badge verifications: 10 per client per minute, and 3 per token in 10 minutes. # /verify/, a page: 30 lookups per client per minute. # A refused request gets HTTP 429 with a Retry-After header. # One request is enough for most agents: summary: https://agenticwebscore.com/llms.txt full-text: https://agenticwebscore.com/llms-full.txt typed-index: https://agenticwebscore.com/index.json skill-packs: https://agenticwebscore.com/.well-known/agent-skills/index.json # Every page in the page registry and every learn article has a markdown twin at # .md, and both honour "Accept: text/markdown". You do not need to parse # HTML to read those. The 3 routes below are not in the registry # and serve HTML only: /scan, /directory, /learn.