agentleFS
Sign inStart free
Works-HQWorks-HQ/ai-brain-starter-pack/company/AGENTS.md

ai-brain-starter-pack / company

AGENTS.md3 starsChanged 3 months agoMITView on GitHub

This is the first file your AI reads on every run. It tells the AI who you are, how you work, and where everything lives. Keep it lean. Fill the placeholders, delete the guidance notes once you have, and add to it as your world gets clearer. CLAUDE.md is canonical. In your deployed folder, point AGENTS.md to it with a relative symlink where supported. Otherwise generate an identical copy and check for drift. Never maintain two hand-edited instruction files. <2-4…

  • 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 31,774 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 repository6 other files it gives its agents.

The file

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

9 rules. About 0 are common, 7 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    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. Very specific:

    Use this as a short map. Keep durable company context in brain/Company.md and context about people in brain/People/.

    line 29
  2. Distinctive:

    The AI drafts, you send: No autonomous sending of email, messages, or posts.

    line 48
  3. Distinctive:

    Capture immediately, file later: Never lose a thought to "I will remember it." Dump it in, sort it after.

    line 50
  4. Distinctive:

    Read the relevant brain file before working on a business or person.

    line 57
  5. Distinctive:

    Read the current operating-system files before changing live state, tasks, decisions, or logs.

    line 58
  6. Very specific:

    When you are unsure, mark it [confirm] rather than stating a guess as fact.

    line 59
  7. Distinctive:

    Never overwrite an existing brain file.

    line 62
  8. Distinctive:

    No emdashes. Use commas, periods, or line breaks.

    line 63
  9. Distinctive:

    If you use Gemini CLI, copy CLAUDE.md to GEMINI.md.

    line 71

≈ “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

Context Root

This is the first file your AI reads on every run. It tells the AI who you are, how you work, and where everything lives. Keep it lean. Fill the placeholders, delete the guidance notes once you have, and add to it as your world gets clearer.

CLAUDE.md is canonical. In your deployed folder, point AGENTS.md to it with a relative symlink where supported. Otherwise generate an identical copy and check for drift. Never maintain two hand-edited instruction files.

Who I am

  • Name: <YOUR NAME>
  • What I do: <ONE LINE: your role and the businesses you run or work across>
  • Where I work from: <machines / OS, if it matters>

<2-4 lines: your story, what you are building, what matters to you. Enough that the AI talks to you as you, not as a generic user.>

How I decide and how I sound

  • Decision style: <fast and decisive / considered and thorough / data first / gut then check>
  • Voice: <how you want drafts to sound: direct, warm, plain, formal. Give an example line if you can.>
  • What I do not want: <corporate filler, hype, hedging, whatever grates on you>

My world

Use this as a short map. Keep durable company context in brain/Company.md and context about people in brain/People/.

  • <BUSINESS / PROJECT>, <one line>, context in brain/Company.md
  • <BUSINESS / PROJECT>, <one line>

How this deployed folder is laid out

  • brain/ holds slow-changing context about the company and its people. Read the relevant file before doing related work. Ignore brain/_source-docs/ on normal runs because it is raw source material, not refined context.
  • operating-system/ holds fast-changing state in current-context.md, one task list in GLOBAL_TASKS.md, and the working files created from operating-system/templates/.
  • skills/ holds the repeatable workflows: debrief, reflect, morning sweep, weekly update, and meeting notes.

These paths are relative to your deployed root folder. They do not point back into the starter-pack repository.

The hard rules (non-negotiable)

  • The AI drafts, you send. No autonomous sending of email, messages, or posts. Ever. Every reply is drafted by the AI and sent by you.
  • Manual first, automate later. Prove a workflow is useful twice before you schedule or automate it.
  • Capture immediately, file later. Never lose a thought to "I will remember it." Dump it in, sort it after.
  • Never overwrite the brain. This system wraps around what you already have. It adds a layer, it never replaces your existing files.

How to work with me

  • Read the relevant brain file before working on a business or person. Do not answer from a blank slate when context exists.
  • Read the current operating-system files before changing live state, tasks, decisions, or logs.
  • When you are unsure, mark it [confirm] rather than stating a guess as fact.
  • Flag anything legal, financial, HR, or otherwise sensitive for my review before filing it anywhere.
  • Show me the plan before writing across any files. Wait for approval, then write only the approved changes.
  • Never overwrite an existing brain file. Update approved facts in place and preserve everything else.
  • No emdashes. Use commas, periods, or line breaks.

A note on tools

This system is tool-agnostic. It is markdown files, a task list, and a rhythm. Any capable AI agent that can read and write files in this folder can run it. Claude Code and Codex are examples, not requirements. Think of the runtime as rented: the model and the interface will keep changing, but this folder is what you own, and it moves with you.

If you use Gemini CLI, copy CLAUDE.md to GEMINI.md.

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.

Check your setup, free