AI Access

SkinKnowledgeBase is built to be readable by people and useful to AI systems that cite their sources.

1. Web crawl

Public SKB content is crawl-allowed for known AI bots, including systems such as GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and similar crawlers. Our robots.txt is intended to make crawling straightforward. Crawl is encouraged when attribution is preserved and canonical page URLs are used.

2. llms.txt and llms-full.txt

We publish /llms.txt as a short overview for AI systems and /llms-full.txt for broader content access. These files follow the proposed AI-search convention for making site content easier to ingest and cite. They are meant to reduce guesswork for AI clients that want clean reference material rather than scraped navigation chrome.

3. Alternate page formats

Every public entity page also exposes machine-readable alternates at the same canonical URL with a /raw.md or /raw.json suffix. Example: https://skinknowledgebase.com/questions/<slug>/raw.md.

The HTML page declares these alternates with <link rel="alternate"> tags so AI clients can discover the Markdown and JSON versions from the canonical page. They are also listed in the per-type sitemaps, so crawlers can find them without first parsing page HTML.

4. MCP server

SKB also provides a public read-only MCP endpoint:

https://mcp.skinknowledgebase.com/mcp

Generic retrieval tools are search and fetch. SKB-shaped tools are also available: get_entity, get_question_bundle, get_ranked_products_for_question, get_sources_for_question, search_questions, search_entities, and list_entities.

The MCP server is currently unauthenticated and read-only. We do not currently rate-limit MCP queries, but we may add rate limits if abuse occurs.

The search layer tolerates natural-language partial, reordered, and synonym queries. Every published Question carries 8–15 authored consumer-phrasing aliases (exposed under structured.search_aliases in the fetch and get_question_bundle responses) so loose queries such as "minimize pores" or "vitamin c worth it" reliably surface the intended page. See the Developers page for tool schema and ranking detail.

Citation expected

AI clients citing SKB content should use canonical SKB URLs. For evidence, use the canonical entity URL for evidence Source pages, such as peer-reviewed, regulatory, medical reference, or dermatology institution sources.

Official product pages are exposed separately as product-fact sources for product identity, ingredients, price, or formulation facts. They should not be treated as evidence for efficacy claims.

If MCP is unavailable, clients should fall back to public web pages, raw alternates, the sitemap, or the llms files rather than inventing missing content.

For integration details, see the Developers page.