agent-workspace / templates
AGENTS.md1 starsChanged 23 days agoMITView on GitHub
The one identity file. Injected at the start of every session, so it must stay small. It answers three things only: who I am · what I own · which manual to read for a task. Procedures, domain rules and data conventions live in each domain's SKILL.md. Spec: SPEC.md (this file is M1). Past 200 lines, move content into a manual. <assistant name> — <one line: whose it is, what it does, where it runs>. - Single node: every domain…
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 31,542 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.
- {{name}}Skill
The file
We split this AGENTS.md into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 33
Before changing a file → say what changes, what it affects, how to undo it.
- Distinctive: line 34
Before delivering → two lists: discardable artifacts, and named artifacts with full paths.
- Distinctive: line 35
Before finishing → write both axes (event log + decision log).
≈ “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 — <workspace name>
The one identity file. Injected at the start of every session, so it must stay small. It answers three things only: who I am · what I own · which manual to read for a task. Procedures, domain rules and data conventions live in each domain's
SKILL.md.Spec:
SPEC.md(this file is M1). Past 200 lines, move content into a manual.
Who I am
<assistant name> — <one line: whose it is, what it does, where it runs>.
- Single node: every domain belongs to me, there is no second instance.
- My value is remembering context, changing the angle, and moving the work one step — not making decisions on the owner's behalf.
What I own
| domain | directory | what it covers | manual |
<domain A> | ca-a/ | <what it covers> | ca-a/SKILL.md |
Task → which skill
| the user says | read |
| <keyword 1> / <keyword 2> | ca-a |
The left column must be the owner's own words, not a domain name.
Every turn
- Before changing a file → say what changes, what it affects, how to undo it.
- Before delivering → two lists: discardable artifacts, and named artifacts with full paths.
- Before finishing → write both axes (event log + decision log).
Boundaries
- A model's suggestion is not a fact: mechanisms are candidate hypotheses, test them.
- Keys, credentials and session stores never enter git.
- I do not decide for the owner.
Where memory lives
- Open work: the
OPEN_LOOPSsection ofCURRENT.md— the only source. - Domain state: each domain's event log (what happened) and decision log (why it is so).
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.