agentleFS
kylehugheskylehughes/distillation-for-agents/AGENTS.md

distillation-for-agents

AGENTS.md1 starsChanged 6 days agoMITView on GitHub

This repository distributes the Distill skill for Claude Code, Codex, and other Agent Skills hosts. Keep one shared skill under skills/distill/. plugin.json is the portable manifest. .codex-plugin/plugin.json supplies the compatible Codex manifest and interface metadata; .claude-plugin/plugin.json declares the Claude plugin. Both marketplace files point to this repository root. Keep identities and release versions synchronized; do not rename the published plugin or skill without an explicit migration. Run python3 scripts/check.py, python3 -m unittest discover -s scripts -p 'test_*.py', and python3 scripts/package.py…

  • 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 31,700 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 repository2 other files it gives its agents.

The file

We split this AGENTS.md into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep the procedure domain-general and proportionate.

    line 9
  2. Very specific:

    When adding or changing a research claim, update references/sources.md with the work, passage, edition, and caveat, and keep source IDs …

    line 10
  3. Very specific:

    Run python3 scripts/check.py, python3 -m unittest discover -s scripts -p 'test_*.py', and python3 scripts/package.py after changes.

    line 16

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

Repository guidance

This repository distributes the Distill skill for Claude Code, Codex, and other Agent Skills hosts. Keep one shared skill under skills/distill/.

Content

  • SKILL.md contains the six-step procedure. The files in references/ (operators, validation, examples, foundations, sources) are loaded only when a task needs them. Keep detailed guidance there rather than duplicating it in the entrypoint.
  • agents/openai.yaml supplies the skill's Codex UI metadata and assets/icon.svg its icon. Keep display_name, short_description, and default_prompt as single-line quoted strings so the checks can read them.
  • Keep the procedure domain-general and proportionate. Simplification must preserve the distinctions, evidence, uncertainty, and claim strength the task requires. Do not turn the skill into a mandatory checklist or promise a uniquely simplest answer.
  • When adding or changing a research claim, update references/sources.md with the work, passage, edition, and caveat, and keep source IDs stable.

Packaging and checks

plugin.json is the portable manifest. .codex-plugin/plugin.json supplies the compatible Codex manifest and interface metadata; .claude-plugin/plugin.json declares the Claude plugin. Both marketplace files point to this repository root. Keep identities and release versions synchronized; do not rename the published plugin or skill without an explicit migration.

Run python3 scripts/check.py, python3 -m unittest discover -s scripts -p 'test_*.py', and python3 scripts/package.py after changes. Validate the Claude manifests with claude plugin validate .claude-plugin/plugin.json and claude plugin validate .claude-plugin/marketplace.json. For runtime checks, use the local marketplace in Codex and confirm skill discovery in both hosts.

Each release ships three archives. distill-skill.zip keeps SKILL.md at its root for compatibility. distill.skill is the Claude app upload format, with the distill/ folder at its root; scripts/check.py enforces that format's frontmatter rules (allowed keys, a kebab-case name, a description of at most 1024 characters without angle brackets). distill-plugin.zip holds the plugin. All three archives must include the license. Package only runtime content; exclude repository instructions, build tooling, local settings, and caches.

Releases

./release.sh --dry-run <version> checks release preconditions without changing files. ./release.sh <version> updates the manifests, validates, commits Release <version>, creates the matching tag, and pushes the main branch and that tag. Run the publishing command only when a release is requested. GitHub Actions validates the tag and builds all three archives before publishing the release.

License

Distributed under the MIT License. Preserve LICENSE in every distribution.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.