# Every surface AI agents need. Automatically. Every SaturnDocs documentation site automatically publishes and maintains the files, protocols, tools, and metadata AI agents need to discover, read, and cite your product. × Agent surfaces Migrate to SaturnDocs and your documentation ships as more than a website. It includes the discovery files, clean source formats, protocols, and metadata that make your product easier for AI agents to find, understand, use, and recommend. _1_You ship_2_We rebuild_3_Agents discover A moving target, handled ## You should not have to track the next agent standard. Agent discovery is changing quickly. Today that means Markdown mirrors, llms.txt, Model Context Protocol, WebMCP, skills, and Agent2Agent. Tomorrow it will include surfaces that do not exist yet. SaturnDocs tracks that work, implements the useful standards, and adds them across customer documentation sites. Your team keeps building the company; the documentation stays legible to the agents your customers use. Alwaysagent-ready - Rebuilt on every deploy - New standards added - Permissions preserved Every site ### llms.txt **A concise documentation index** A generated list of canonical pages with titles, descriptions, and direct URLs. It gives an agent a quick map of what the documentation contains. Why it matters An agent can find the right page without crawling the visual site or guessing from navigation labels. `/llms.txt` Every site ### llms-full.txt **The complete public corpus in one file** A generated text file containing the clean Markdown from every page the requesting reader is allowed to access. Why it matters An agent can load the full documentation into context when a task spans many pages. `/llms-full.txt` Every site ### Agent Skills index **A directory of instructions for agents** A machine-readable index that names each published skill, explains when to use it, points to its instructions, and verifies those instructions with a digest. Why it matters Compatible agents can discover the site skill automatically and confirm that the instructions have not drifted. `/.well-known/agent-skills/index.json` Every site ### Site skill **Instructions for using this documentation** A generated skill that tells an agent how to choose among llms.txt, full-site text, page Markdown, Model Context Protocol tools, and authenticated content. Why it matters The agent learns the best way to read the site instead of improvising a fragile scraping strategy. `/.well-known/agent-skills/{skill}/SKILL.md` Every page ### Markdown mirrors **One clean Markdown twin per page** Every documentation page publishes a Markdown representation without navigation, decorative interface elements, or duplicated reader chrome. Why it matters Agents receive the actual documentation in a compact format that is easier to quote, reason over, and cite. `/guide.md` Every site ### MCP server **Model Context Protocol tools** A read-only Streamable HTTP server with tools to search, list, fetch, grep, and ask questions across the published documentation. Why it matters An agent can retrieve only the evidence it needs without leaving its coding or work environment. `/mcp` Access-aware ### WebMCP **Documentation tools inside the browser** Compatible browser agents can discover the documentation tools directly from the page they are visiting. Public pages enable the bridge automatically; private access remains protected. Why it matters A browser agent can search and read the docs as tools instead of trying to interpret the visual interface. `Registered on documentation pages` Public sites ### A2A endpoint **Cited answers over Agent2Agent** A JSON-RPC endpoint that answers questions from the published corpus and returns citations to the supporting documentation pages. Why it matters Another agent can ask the documentation agent directly and preserve the source trail in its own workflow. `/a2a` Every site ### MCP server card **The MCP connection contract** A machine-readable description of the protocol version, transport, endpoint, authentication posture, capabilities, and exact tool schemas. Why it matters Clients can configure themselves from a declared contract instead of relying on setup instructions copied by hand. `/.well-known/mcp/server-card.json` Public sites ### A2A agent card **The documentation agent identity** A capability card that advertises the Agent2Agent endpoint, supported input and output modes, protocol version, and documentation question-answering skill. Why it matters Other agents can determine what the documentation agent does and how to call it before sending a request. `/.well-known/agent-card.json` Every site ### API catalog **A catalog of callable interfaces** A standards-based linkset that advertises the site’s Model Context Protocol metadata and each public OpenAPI description published with the docs. Why it matters Agents have one predictable place to discover both product APIs and documentation tools. `/.well-known/api-catalog` API docs ### Published OpenAPI **The source API contract** Public OpenAPI descriptions are published in a stable machine-readable location and connected to their rendered API reference. Why it matters An agent can inspect the exact API schema behind the guide instead of reconstructing endpoints from prose. `/api-reference/{name}.openapi.json` Every site ### robots.txt and content signals **Crawler and usage guidance** A generated crawler policy publishes the sitemap and the site’s search, AI-input, and AI-training content signals. Private sites fail closed. Why it matters Automated readers receive a current, explicit policy instead of inferring one from inconsistent headers or stale files. `/robots.txt` Every site ### Authentication guide **Machine-readable access instructions** A plain-language guide explains whether the site is public, private, or mixed, and what an unauthenticated agent can read. Why it matters Agents know when content requires a signed-in reader and never mistake a public subset for the complete corpus. `/auth.md` Every site ### Sitemaps and canonical metadata **Stable page discovery and identity** Generated sitemaps, canonical URLs, and alternate Markdown links keep every representation tied to the same authoritative page. Why it matters Search systems and agents can discover the full site without creating duplicate or conflicting page identities. `/sitemap-index.xml` Every page ### Structured page data **Machine-readable page meaning** Each reader page describes itself as a technical article and includes its site relationship, canonical URL, modification date, and breadcrumb trail. Why it matters Agents can understand page context and hierarchy without deriving them from visual layout. `JSON-LD in each page` Continuous coverage ## New agent standards become part of your docs automatically. When a meaningful discovery standard or agent interface emerges, our team evaluates it, builds it into SaturnDocs, and rolls it out to every applicable documentation site. There is no separate plugin to choose, generator to maintain, or standards project for your product team to staff. Access-aware surfaces follow the site’s reader permissions. Private documentation stays private. 1. 01 **A new standard emerges** We track the formats and protocols agents are actually adopting. 2. 02 **SaturnDocs adds and tests it** We implement discovery, generation, access behavior, and deployment checks once. 3. 03 **Your documentation receives it** The new surface rolls out with the platform and stays synchronized from then on. One source, every interface ## Change a page once. Every surface follows. Each deployment renders the human site, derives the clean Markdown, rebuilds the indexes and skills, refreshes the protocol contracts, and publishes the current corpus behind the live endpoints. Access rules are applied to every representation, so agent surfaces do not become stale copies or permission bypasses. SourceHTMLMarkdownDiscoveryProtocols Agent-first by default ## Move once. Stay current automatically. SaturnDocs keeps every useful agent surface synchronized with your documentation as the standards evolve. [Send us your docs](https://saturndocs.com/contact/?intent=docs) Canonical URL: https://saturndocs.com/agent-surfaces/