agents.txt is a proposed discovery file for publishing a site's agent-facing capabilities. It is designed for AI agents that may need to discover supported protocols, payment mechanisms, authorization schemes, MCP endpoints, skills, A2A agent cards, or other machine-readable interaction surfaces.
Where llms.txt helps an agent find content to read, agents.txt helps an agent understand what it can do with a site. It is typically paired with a structured agents.json file for richer machine parsing.
Examples of capabilities that may be declared include:
- MCP server endpoints;
- Agent2Agent Protocol (A2A) Agent Card URLs;
- skill package URLs;
- authorization discovery such as Agent Auth Protocol;
- agent payment support such as Agent Payments Protocol (AP2); and
- commerce or interaction profiles.
The file is not a trust guarantee. Agents still need to validate endpoints, authenticate properly, enforce agent policy, and ask for human approval when actions are consequential.
The agents.txt standard draft defines directives for payments, authorization, MCP, skills, A2A, UCP, and WebMCP discovery.
The LLM Knowledge Base is a collection of bite-sized explanations for commonly used terms and abbreviations related to Large Language Models and Generative AI.
It's an educational resource that helps you stay up-to-date with the latest developments in AI research and its applications.