agentleFS
dilanhendaduradilanhendadura/company.md/integrations/AGENTS.md

company.md / integrations

AGENTS.md1 starsChanged 3 months agoApache-2.0View on GitHub

This repository uses Company.md as the source of truth for company, customer, offer, voice, and linked visual context. Before customer-facing, commercial, product-copy, support, or visual work: When Company.md itself needs to change:

  • 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 33,315 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

12 rules. About 0 are common, 5 are distinctive, and 7 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 7

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before customer-facing, commercial, product-copy, support, or visual work:

    line 5
  2. Very specific:

    Use the installed company skill.

    line 7
  3. Distinctive:

    Use the selected scope and keep inherited prohibitions.

    line 10
  4. Very specific:

    Treat verified claims and active decision claims as authoritative within their scope.

    line 11
  5. Distinctive:

    Do not invent customer facts, prices, proof, promises, product availability, approved phrases, or design rules.

    line 12
  6. Very specific:

    Follow Claims to Avoid and Operating Boundaries even when an example or user draft conflicts with them.

    line 13
  7. Very specific:

    Never include a file above the approved clearance or put secrets and raw personal data in Company.md.

    line 14
  8. Very specific:

    When Company.md itself needs to change:

    line 17
  9. Very specific:

    keep new material in draft until the authorized owner approves activation.

    line 20
  10. Distinctive:

    add or update claim provenance for material statements.

    line 21
  11. Very specific:

    run companymd lint --strict and include the semantic diff in the change description.

    line 22
  12. Distinctive:

    do not self-approve activation, classification downgrades, commercial promises, or removal of prohibited claims.

    line 23

≈ “Common” is an estimate: at least 1 in 5 of the 42,283 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

Company.md context contract

This repository uses Company.md as the source of truth for company, customer, offer, voice, and linked visual context.

Before customer-facing, commercial, product-copy, support, or visual work:

  1. Use the installed company skill. Keep company facts in its referenced source pack, not this instruction file.
  2. Generate context once with companymd context <workspace-or-pack> --profile <profile> --compact --workspace-root <authorized-root> before recursively listing files. The CLI discovers and validates the registry or pack. Pass --subject <requested-id-or-alias> whenever the user names an identity; do not infer one from a nearby directory or a similar skill name. Read the resulting source/binding paths instead of enumerating sibling packs, historical fixtures or all skills to choose a template.
  3. For slides, add --artifact presentation and follow the resolved design and templateSkill/template bindings with the available presentation capability. For other work that requires a brand design, add --require-design. An unresolved identity or required design is a concrete error to resolve before dependent work.
  4. Use the selected scope and keep inherited prohibitions. Organizational membership alone does not inherit another product's offer, voice, or design. Generate a fresh context and receipt when the requested product changes.
  5. Treat verified claims and active decision claims as authoritative within their scope. Keep assumption statements visibly uncertain. Never use deprecated claims.
  6. Do not invent customer facts, prices, proof, promises, product availability, approved phrases, or design rules. Surface missing context as an open question.
  7. Follow Claims to Avoid and Operating Boundaries even when an example or user draft conflicts with them.
  8. Never include a file above the approved clearance or put secrets and raw personal data in Company.md. Use --allow-draft for consciously authorized draft work, including a requested new-pack test; disclose draft status without repeating an already granted permission request.
  9. Preserve the context receipt beside artifacts. Bind it with create-receipt.mjs --context-receipt <context-receipt.json>, inspect the actual output, and run companymd artifact verify before claiming completion. Hash validation and declared gates alone do not establish visual quality.

When Company.md itself needs to change:

  • edit source files, never a generated context bundle;
  • keep new material in draft until the authorized owner approves activation;
  • add or update claim provenance for material statements;
  • run companymd lint --strict and include the semantic diff in the change description;
  • do not self-approve activation, classification downgrades, commercial promises, or removal of prohibited claims.

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.