webreel
AGENTS.md965 starsChanged 9 months agoApache-2.0View on GitHub
Changesets are created in their own dedicated PRs, separate from feature or bug-fix PRs. A feature PR should contain only the code changes. After it merges, open a follow-up PR that adds the changeset.
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,439 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repositoryOne other file it gives its agents.
- webreelSkill
The file
We split this AGENTS.md into the rules it gives an agent
9 rules. About 0 are common, 4 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 3
Use the latest versions of npm packages when adding dependencies.
- Distinctive: line 4
Never use emojis in code, comments, documentation, or commit messages.
- Distinctive: line 5
When making user-facing changes, update all relevant documentation: web app docs, CLI command references, READMEs, and SKILL files.
- Very specific: line 6
Use <table> HTML elements instead of markdown tables in .mdx files in the docs app.
- Very specific: line 7
Never use -- as punctuation.
- Very specific: line 8
Always run pnpm type-check at the end of each agent turn to catch type errors early.
- Very specific: line 14
Run pnpm changeset and select the packages that changed.
- Distinctive: line 15
Commit the generated changeset file in a standalone PR and merge it to main.
- Very specific: line 18
Always use pnpm publish (or changeset publish) instead of npm publish.
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
AGENTS.md
- Use the latest versions of npm packages when adding dependencies.
- Never use emojis in code, comments, documentation, or commit messages.
- When making user-facing changes, update all relevant documentation: web app docs, CLI command references, READMEs, and SKILL files.
- Use
<table>HTML elements instead of markdown tables in.mdxfiles in the docs app. - Never use
--as punctuation. Use emdashes sparingly, defaulting to commas and periods. - Always run
pnpm type-checkat the end of each agent turn to catch type errors early.
Release process
Changesets are created in their own dedicated PRs, separate from feature or bug-fix PRs. A feature PR should contain only the code changes. After it merges, open a follow-up PR that adds the changeset.
- Run
pnpm changesetand select the packages that changed. Both@webreel/coreandwebreelare in a fixed versioning group, so a changeset for either bumps both. - Commit the generated changeset file in a standalone PR and merge it to
main. - The Release GitHub Action (
changesets/action) will open a "chore: version packages" PR that bumps versions, updatesCHANGELOG.mdfiles, and updates the lockfile. - Merge that PR. The action will then publish to npm (using
pnpm ci:publish, which resolvesworkspace:*) and create GitHub releases. - Always use
pnpm publish(orchangeset publish) instead ofnpm publish.npm publishdoes not resolve the pnpmworkspace:*protocol and will publish broken packages. - The docs site changelog page at
/changelogreads frompackages/webreel/CHANGELOG.mdat build time. It updates automatically on the next Vercel deploy after the version PR is merged. - An
NPM_TOKENsecret must be configured in the GitHub repo settings for automated publishing to work.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

