AGENTS.md vs CLAUDE.md
AGENTS.md vs CLAUDE.md: side-by-side comparison of two file-backed / editor paradigms systems — architecture, taxonomy, license, pricing, MCP/A2A support, and direct edges.
Cost & capability
| AGENTS.md | CLAUDE.md | |
|---|---|---|
| Use cases | Code Generation Focused, Scoped Agentic | Code Generation Focused, Scoped Agentic, Long Running Session |
Where they differ (4)
Rows where both sides have data and the values disagree — the shortlist of dimensions that actually distinguish these two systems.
| AGENTS.md | CLAUDE.md | |
|---|---|---|
| Use cases | Code Generation Focused, Scoped Agentic | Code Generation Focused, Scoped Agentic, Long Running Session |
| Type | Cross-tool open standard | Git-backed persistent instructions |
| Created | 2025-05 (introduced with OpenAI Codex CLI research preview May 16 2025; donated to Linux Foundation AAIF Dec 2025) | 2024-10 (first appeared in Claude Code public release; Claude Code beta May 2024; GA Oct 2024) |
| Backend storage | file (AGENTS.md) | file (CLAUDE.md) |
At a glance
| AGENTS.md | CLAUDE.md | |
|---|---|---|
| Section | File-backed / editor paradigms | File-backed / editor paradigms |
| Tier | T1 | T1 |
| Type | Cross-tool open standard | Git-backed persistent instructions |
| Created | 2025-05 (introduced with OpenAI Codex CLI research preview May 16 2025; donated to Linux Foundation AAIF Dec 2025) | 2024-10 (first appeared in Claude Code public release; Claude Code beta May 2024; GA Oct 2024) |
| Backend storage | file (AGENTS.md) | file (CLAUDE.md) |
| Optimised for | simplicity / git-versioned / human-readable | simplicity / git-versioned / human-readable |
| Anti-fit | not for non-coding / non-IDE use cases | not for non-coding / non-IDE use cases |
Taxonomy
| Axis | AGENTS.md | CLAUDE.md |
|---|---|---|
| storage | file | file |
| retrieval | injection | injection |
| persistence | cross-session | cross-session |
| update | overwrite | overwrite |
| unit | document | document |
| governance | user-controllable | user-controllable |
| conflict | manual | manual |
Pros & cons
AGENTS.md
Pros: Cross-vendor convention — the closest thing to a universal agent memory file format; works across Claude Code, Cursor, and Codex CLI.
Cons: Adoption is informal (no spec authority); agent-by-agent interpretation differences mean behavior isn't fully portable.
CLAUDE.md
Pros: Plain-text + git-versioned + human-reviewable — the highest-trust memory format with zero infra; works in any editor or terminal.
Cons: Manual curation required; doesn't scale beyond ~10–20 KB before context bloat; no semantic recall, only injection-by-default.