pi-public / pi
AGENTS.md1 starsChanged 4 months agoMITView on GitHub
Hello I am bnema and you are my coding partner for this session. Follow these small rules so our work stays clean, thoughtful, and safe.
We split this AGENTS.md into the rules it gives an agent
16 rules. About 0 are common, 14 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈14
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
- Distinctive: line 5
Keep implementation ownership with the main agent or the developer.
- Distinctive: line 6
Keep architecture decisions and product trade-offs under direct control.
- Distinctive: line 11
Check the current branch before editing.
- Distinctive: line 12
Do not work directly on main.
- Distinctive: line 13
Verify unfamiliar APIs in docs, skills, Context7, or installed source.
- Distinctive: line 14
Prefer source-backed implementation over memory.
- Very specific: line 16
Do not keep old code as legacy unless explicitly requested.
- Distinctive: line 17
Always take some time to explore the code and understand the architecture before implementing anything.
- Distinctive: line 21
Use subagents for focused research, review, verification, or when the work is boring and you don't want to do it yourself.
- Distinctive: line 25
After each completed phase, plan, or significant change, request an appropriate review.
- Distinctive: line 29
Never leak personal information in output, code, tests, docs, or commit messages.
- Distinctive: line 30
Use generic placeholders for paths, names, emails, and other personal details.
- Distinctive: line 34
Always sign the commits or the tags.
- Very specific: line 36
Avoid review-process commit messages like fixing review comments.
- Distinctive: line 38
Keep documentation concise and focused on the key points.
- Distinctive: line 39
Write documentation, README files, Markdown guides, and user-facing prose from the present product state.
≈ “Common” is an estimate: at least 1 in 5 of the 36,368 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
Hello I am bnema and you are my coding partner for this session. Follow these small rules so our work stays clean, thoughtful, and safe.
Collaboration style
- Keep implementation ownership with the main agent or the developer.
- Keep architecture decisions and product trade-offs under direct control.
- Feel free to propose a refactor if it makes the code simpler, clearer, or easier to maintain.
Before implementing
- Check the current branch before editing.
- Do not work directly on
main; create a working branch first when the task is clear. - Verify unfamiliar APIs in docs, skills, Context7, or installed source.
- Prefer source-backed implementation over memory.
- For
v0.xprojects, prefer clean breaking changes over compatibility layers. - Do not keep old code as
legacyunless explicitly requested. - Always take some time to explore the code and understand the architecture before implementing anything.
Delegation and review
- Use subagents for focused research, review, verification, or when the work is boring and you don't want to do it yourself.
- List and delegate to the narrowest specialized subagent.
- Give subagents short, complete briefs.
- Review and verify subagent work before integrating it.
- After each completed phase, plan, or significant change, request an appropriate review.
Privacy and safety
- Never leak personal information in output, code, tests, docs, or commit messages.
- Use generic placeholders for paths, names, emails, and other personal details.
Commits and documentation
- Always sign the commits or the tags.
- Make commit messages describe what changed.
- Avoid review-process commit messages like
fixing review comments. - Format commits as
type(scope): description. - Keep documentation concise and focused on the key points.
- Write documentation, README files, Markdown guides, and user-facing prose from the present product state. Do not frame docs around recent changes, migration history, or what was just implemented unless the document is explicitly a changelog, migration guide, or release note.
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 28,693 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 repository22 other files it gives its agents.
- agent-workflowsSkill
- architecture-securitySkill
- code-review-releaseSkill
- documentation-passSkill
- exploring-codebasesSkill
- find-skillsSkill
- frontend-webSkill
- githubSkill
- grill-meSkill
- grill-with-docsSkill
- improve-codebase-architectureSkill
- improveSkill
- product-design-writingSkill
- project-backlogSkill
- prototypeSkill
- red-teaming-decisionsSkill
- systems-backendSkill
- tddSkill
- testing-debugging-verificationSkill
- to-issuesSkill
- to-prdSkill
- triageSkill
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
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.

