Cursor rules vs AGENTS.md
The same 6 criteria for both, with the page each figure came from and the day it was read.
Which one, and when
On where the file lives, Cursor rules is .cursor/rules/*.mdc and AGENTS.md is AGENTS.md at the repository root. On per-directory files, Cursor rules is Folders inside .cursor/rules, for organisation and AGENTS.md is Yes, one per package. On path-scoped rules, Cursor rules is Yes, globs, description or @-mention and AGENTS.md does not publish it. On what wins on a conflict, Cursor rules is Team rules take precedence and AGENTS.md is Nearest file wins; the chat prompt beats all. On published size guidance, Cursor rules is Under 500 lines per rule and AGENTS.md does not publish it. On what its own docs say about rival formats, Cursor rules is Names AGENTS.md as the plain-markdown option and AGENTS.md is Documents migrating in, by rename and symlink. RuleStack sells neither of these and takes no referral fee on either, it scores them. Every figure was read off the named company's own page on or since 2026-09-13; the URLs and the dates are at the foot of this page, and a job re-fetches each one on a schedule and fails when a figure is no longer on the page it came from.
Cursor rules vs AGENTS.md, row by row
Read 13 September 2026Every figure was read off the named company's own page on or since 13 September 2026. A job re-fetches each URL on a schedule and fails when a figure is no longer on the page it came from.
One paragraph per format, in its own terms
Pick Cursor rules when
You want a rule that only loads when it is relevant, and you want the agent, not a glob you maintained by hand, to make that call. Cursor is the only one of the five whose docs publish the full decision table for when a rule is included: `alwaysApply: true` always, a `globs` pattern on a file match, a `description` alone when "agent reads the description and pulls the rule in when relevant", and nothing at all until you @-mention it. Nobody else documents a description-triggered rule, and on a large repository that is the difference between five rules in context and fifty.
Pick AGENTS.md when
More than one agent touches the repository, or you do not know yet which one will. It is the only format of the five that is not owned by the tool that reads it, "AGENTS.md is now stewarded by the Agentic AI Foundation under the Linux Foundation", and the only one whose own page lists the other vendors' agents as readers rather than as rivals. It is also the smallest thing to write: no frontmatter, no extension rules, no directory to create. "Are there required fields? No. AGENTS.md is just standard markdown."
One question, one answer, nothing collapsed
Cursor rules vs AGENTS.md, where the file lives?
Cursor rules: .cursor/rules/*.mdc. "Project rules live in .cursor/rules as .mdc files and are version-controlled." A plain .md file in that directory is ignored: "api-guidelines.md # ignored (wrong extension)". AGENTS.md: AGENTS.md at the repository root. "Create an AGENTS.md file at the root of the repository. Most coding agents can even scaffold one for you if you ask nicely.". Read from cursor.com and agents.md on 2026-09-13.
Cursor rules vs AGENTS.md, per-directory files?
Cursor rules: Folders inside .cursor/rules, for organisation. The documented layout puts subfolders inside the one rules directory, "frontend/ # organize rules in folders", rather than a rules directory beside each package. For per-directory files Cursor's own page points elsewhere: "Cursor supports AGENTS.md in the project root and subdirectories.". AGENTS.md: Yes, one per package. "Large monorepo? Use nested AGENTS.md files for subprojects … agents automatically read the nearest file in the directory tree, so the closest one takes precedence." The page cites the main OpenAI repository as carrying 88 of them. Read from cursor.com and agents.md on 2026-09-13. Per-directory files here means whether the format documents putting a second file next to the code it governs, rather than one file at the repository root.
Cursor rules vs AGENTS.md, path-scoped rules?
Cursor rules: Yes, globs, description or @-mention. Four documented rule types. "Apply to specific files When file matches a specified pattern", and the frontmatter table spells out what each combination of alwaysApply, description and globs does. AGENTS.md does not state it on agents.md. The format carries no frontmatter and no pattern-matching mechanism; scope is decided by where the file sits. Read on agents.md, which states "Are there required fields? No. AGENTS.md is just standard markdown. Use any headings you like" and documents no glob, path or conditional field. Read from cursor.com and agents.md on 2026-09-13. Path-scoped rules here means whether a rule can be limited to files matching a pattern instead of loading into every request, which is the mechanism that decides how much of your context window the file costs.
When should I pick Cursor rules, and when AGENTS.md?
Cursor rules, You want a rule that only loads when it is relevant, and you want the agent, not a glob you maintained by hand, to make that call. Cursor is the only one of the five whose docs publish the full decision table for when a rule is included: `alwaysApply: true` always, a `globs` pattern on a file match, a `description` alone when "agent reads the description and pulls the rule in when relevant", and nothing at all until you @-mention it. Nobody else documents a description-triggered rule, and on a large repository that is the difference between five rules in context and fifty. AGENTS.md, More than one agent touches the repository, or you do not know yet which one will. It is the only format of the five that is not owned by the tool that reads it, "AGENTS.md is now stewarded by the Agentic AI Foundation under the Linux Foundation", and the only one whose own page lists the other vendors' agents as readers rather than as rivals. It is also the smallest thing to write: no frontmatter, no extension rules, no directory to create. "Are there required fields? No. AGENTS.md is just standard markdown.".