# Discovery & Declaration

> how a site announces itself to agents

_6 of 28 · [all The Agent Protocol Atlas](/protocols)_

- [llms.txt](/protocols/discovery/llms-txt) — A markdown file at the domain root that gives language models a curated map of a site's most important content.
- [NLWeb](/protocols/discovery/nlweb) — Turns a website into a conversational, queryable endpoint by combining Schema.org data, a vector store and an LLM — and every NLWeb endpoint is also an MCP server.
- [agents.json](/protocols/discovery/agents-json) — A manifest that describes a site's APIs and workflows in a form agents can discover and execute, typically served from /.well-known/.
- [agents.txt](/protocols/discovery/agents-txt) — A root-level file that declares a site's identity, terms of use for AI agents, service catalog and agentic endpoints in a machine-readable form.
- [DNS-AID](/protocols/discovery/dns-aid) — Lets organizations publish and discover AI agents through standard DNS records — a naming convention over SVCB/TXT/TLSA records, signed with DNSSEC, with no new record types or servers.
- [schema.org](/protocols/discovery/schema-org) — The shared structured-data vocabulary (JSON-LD) that gives agents machine-readable entities and relationships to read off a page — the noun layer the agentic web is built on.