# Agent Skills vs AGENTS.md

The map lists them as complementary: sites and agents commonly implement both. They answer different questions: Agent Skills is a step 5 (Act) standard, AGENTS.md step 4 (Read), so they are not substitutes.

[5 Act](/category/act/) [Agent Skills](/protocols/anthropic-agent-skills-format/) · [4 Read](/category/read/) [AGENTS.md](/protocols/agents-md/)

|  | [Agent Skills](/protocols/anthropic-agent-skills-format/) | [AGENTS.md](/protocols/agents-md/) |
| --- | --- | --- |
| Lifecycle step | 5 Act: Do something for me | 4 Read: Give me the content |
| Answers | How can reusable, on-demand procedural instructions and bundled resources be packaged so any compatible agent can load and run them? | What build, test, and convention instructions should a coding agent follow in this repository? |
| Governed by | Originated by Anthropic and donated on 22 October 2026 to the Linux Foundation's Agentic AI Foundation (AAIF); developed in the open at github.com/agentskills/agentskills alongside MCP and A2A | Agentic AI Foundation (AAIF), a Linux Foundation project |
| Status (issuer's words) | Active open format under the Linux Foundation Agentic AI Foundation (AAIF), adopted across 35+ agent products | "An open format... regardless of which coding agent you use"; explicitly not proprietary to any one tool |
| Version | unversioned living spec (agentskills.io/specification) | current site guidance (living document) (2026-09) |
| Form / transport | document format | well-known file |
| Found at | /.well-known/agent-skills/index.json | AGENTS.md |
| Implementation | multi-implementation | multi-implementation — agents.md (site claims over 60,000 AGENTS.md files across open-source GitHub repositories, via a path:AGENTS.md NOT is:fork NOT is:archived code search) |
| Evidence | 3 verified | 2 reported, 1 verified |
| Primary use case | Describe an agent's capabilities | Serve readable content to agents |
| Last verified | 2026-10-23 (current) | 2026-10-23 (current) |
| Specification | [https://agentskills.io/specification](https://agentskills.io/specification) | [https://agents.md](https://agents.md) |

## When to use which

- Choose Agent Skills when your question is “How can reusable, on-demand procedural instructions and bundled resources be packaged so any compatible agent can load and run them?” and you accept its maturity: status "Active open format under the Linux Foundation Agentic AI Foundation (AAIF), adopted across 35+ agent products", implementation multi-implementation.
- Choose AGENTS.md when your question is “What build, test, and convention instructions should a coding agent follow in this repository?” and you accept its maturity: status ""An open format... regardless of which coding agent you use"; explicitly not proprietary to any one tool", implementation multi-implementation (agents.md (site claims over 60,000 AGENTS.md files across open-source GitHub repositories, via a path:AGENTS.md NOT is:fork NOT is:archived code search)).

Guidance is derived from the two records (question, status, implementation, prerequisites, pairings), not from opinion. Each record's page carries its numbered references.

## Read the full articles

- [Agent Skills](/protocols/anthropic-agent-skills-format/): How can reusable, on-demand procedural instructions and bundled resources be packaged so any compatible agent can load and run them?
- [AGENTS.md](/protocols/agents-md/): What build, test, and convention instructions should a coding agent follow in this repository?

Ask the MCP server to compare any two: the compare-protocols prompt, or get on both ids. [Connect](/connect/).

[anthropic-agent-skills-format.json](/data/protocols/anthropic-agent-skills-format.json) · [agents-md.json](/data/protocols/agents-md.json) · [Markdown](/compare/anthropic-agent-skills-format-vs-agents-md/index.md)