# agents.txt Standard

> The open specification for AI agent capability declarations. Layer 4 of the agent-readiness stack.

## Specification
- [agents.txt Standard v1.0](https://agents-txt.com): The open specification for AI agent capability declarations. Covers file format, directives, discovery, payment protocols (x402, MPP), authorization (agent-auth), MCP endpoint declaration, and skills. Layer 4 of the agent-readiness stack.
- [agents.txt Standard v1.0 (full text)](https://agents-txt.com/spec): The complete text of the v1.0 specification rendered as a single page: file format, directives, discovery, the agents.json schema, the four capability blocks (Payments, Authorization, MCP, Skills) plus A2A and UCP, §4.5 serving requirements, and the registries defined in §17.
- [agents.txt Registry](https://agents-txt.com/registry): Live surface for the two registries defined in §17 of the spec: the directive names registered for use in agents.txt, and the per-protocol object shapes registered for use in agents.json. Authoritative between spec versions; the next spec version absorbs accumulated changes.
- [Audit a site against the agents.txt spec](https://agents-txt.com/audit): Audits any live site against the agents.txt specification: validates /agents.txt and /agents.json against §3–§11, the §5 schema, §4.5 serving headers, and cross-file consistency, and reports an agent-readiness score. Runs in the browser or via the audit_site MCP tool.
- [About agents.txt](https://agents-txt.com/about): Working group, governance, licensing (CC0 spec / Apache 2.0 code), adoption paths, and contact channels. Trust anchor page covering the questions agents and developers ask before adopting the spec.
- [agents.txt vs other agent-readiness standards](https://agents-txt.com/compare): Side-by-side comparison of agents.txt against MCP, A2A, UCP, OpenAPI, OAuth 2.0, llms.txt, robots.txt, ACP, WebMCP, and Agent Skills Discovery. agents.txt is the discovery layer that lets agents find which capabilities a site exposes before negotiating any single protocol.
- [AGENTS.md (agent instructions)](https://agents-txt.com/AGENTS.md): When-to-use instructions for AI agents working against this repository: scope of the spec, boundary with the herald toolkit, where each capability lives. Same content as the AGENTS.md at the repo root, served at the agents.md convention path.

## Demos
- [Demos index](https://agents-txt.com/demo): Live demonstrations of every capability the spec advertises.
- [Agent Auth demo](https://agents-txt.com/demo/auth): 7-step Ed25519 + JWT handshake against the agent-auth Cloudflare Worker.
- [auth.md demo](https://agents-txt.com/demo/authmd): Agentic registration walkthrough: PRM discovery, AS metadata with agent_auth block, anonymous registration, optional OTP claim ceremony, credential use.
- [MCP demo](https://agents-txt.com/demo/mcp): Streamable HTTP MCP session: initialize, tools/list, get_spec, validators.
- [Payments demo](https://agents-txt.com/demo/payments): Announcement-then-wire walkthrough: reads agents.json, then fetches the synthetic /x402 gated route to show the payTo recipient in a real 402.
- [MPP demo](https://agents-txt.com/demo/mpp): Announcement-then-wire walkthrough: reads agents.json, then probes /mpp for a real 402 + WWW-Authenticate: Payment challenge composed by mppx. Tempo and/or Stripe methods activate per the credentials configured on the worker.
- [A2A discovery demo](https://agents-txt.com/demo/a2a): A2A AgentCard discovery flow: reads the A2A: directive from agents.txt and the a2a[] block from agents.json, fetches the declared AgentCard, parses its capabilities and skills, then confirms cross-file consistency via the MCP audit_site tool.
- [UCP discovery demo](https://agents-txt.com/demo/ucp): UCP profile discovery flow: reads the UCP: directive from agents.txt and the ucp[] block from agents.json, fetches the declared profile, parses its services / capabilities / payment_handlers (including the AP2 mandate extension) and signing keys, then confirms cross-file consistency via the MCP audit_site tool.
- [Skills demo](https://agents-txt.com/demo/skills): agents.json skills index → MCP get_skill → installable skill package.
- [Discovery & content layers demo](https://agents-txt.com/demo/llms): Fetches every discovery and content file served by this site: /robots.txt (Layer 1), /sitemap.xml (Layer 2), /llms.txt and /llms-full.txt (Layer 3), plus a §4.5 headers check that confirms /agents.txt and /agents.json serve with the right Content-Type, CORS, and Cache-Control.
- [File Generator](https://agents-txt.com/demo/generate): Browser-only configurator that emits agents.txt + agents.json from form input.