AGENTS.md vs GitHub Copilot custom instructions

AGENTS.md vs GitHub Copilot custom instructions: side-by-side comparison of two file-backed / editor paradigms systems — architecture, taxonomy, license, pricing, MCP/A2A support, and direct edges.

AGENTS.md · GitHub Copilot custom instructions

Cost & capability

AGENTS.mdGitHub Copilot custom instructions
Cost tiermid
Use casesCode Generation Focused, Scoped AgenticCode Generation Focused, Scoped Agentic

Where they differ (3)

Rows where both sides have data and the values disagree — the shortlist of dimensions that actually distinguish these two systems.

AGENTS.mdGitHub Copilot custom instructions
TypeCross-tool open standard.github/copilot-instructions.md + path-scoped
Created2025-05 (introduced with OpenAI Codex CLI research preview May 16 2025; donated to Linux Foundation AAIF Dec 2025)2024 (public preview launched October 2024 in VS Code; .github/copilot-instructions.md format)
Backend storagefile (AGENTS.md)file (.github/copilot-instructions.md)

At a glance

AGENTS.mdGitHub Copilot custom instructions
SectionFile-backed / editor paradigms File-backed / editor paradigms
TierT1 T1
TypeCross-tool open standard .github/copilot-instructions.md + path-scoped
Created2025-05 (introduced with OpenAI Codex CLI research preview May 16 2025; donated to Linux Foundation AAIF Dec 2025) 2024 (public preview launched October 2024 in VS Code; .github/copilot-instructions.md format)
GitHub 0★ no data — repo unreachable
Backend storagefile (AGENTS.md) file (.github/copilot-instructions.md)
Multi-tenancy hard-isolation (per repo/org)
Optimised forsimplicity / git-versioned / human-readable simplicity / git-versioned / human-readable
Anti-fitnot for non-coding / non-IDE use cases not for non-coding / non-IDE use cases

Taxonomy

AxisAGENTS.mdGitHub Copilot custom instructions
storagefilefile
retrievalinjectioninjection
persistencecross-sessioncross-session
updateoverwriteoverwrite
unitdocumentdocument
governanceuser-controllableuser-controllable
conflictmanualmanual

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.

GitHub Copilot custom instructions

Pros: Reaches the largest installed base of AI-using developers; integrated where developers already work; backed by GitHub's persistence.

Cons: Repository-scoped instructions only — no persistent learning; less expressive than Cursor Rules' selector model.

Rows last verified 2026-05-14 / 2026-05-14. Data is CC-BY-4.0 — see how to read this.