nurettincobannurettincoban/ai-prd-workflow/AGENTS.md

ai-prd-workflow

AGENTS.md300 starsChanged 42 days agoMITView on GitHub

This repository is a set of prompts. The prompt files at the repo root are the product; everything else exists to ship them and to check them. Read CONTRIBUTING.md for the full rules; these are the ones an AI agent most often gets wrong. Before committing, run: When writing prompt text: keep it tool-agnostic and concrete, and say why a rule exists when the reason is not obvious — the existing prompts do this, and the reasons are what stop…

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 repository12 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- create-prdSkill
- document-existingSkill
- extract-featuresSkill
- generate-rfcsSkill
- generate-rulesSkill
- implement-rfcSkill
- manage-changesSkill
- review-rfcSkill
- test-strategySkill
- verify-prdSkill
- workflow-statusSkill

The file

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

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

≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
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:
Contracts move with the prompts: If a prompt starts or stops reading or writing an artifact (PRD.md, FEATURES.md, RFCS.md, ...), update the …
line 7
Distinctive:
When writing prompt text: keep it tool-agnostic and concrete, and say why a rule exists when the reason is not obvious — the existing …
line 20

≈ “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

Working on ai-prd-workflow

Edit the root prompts, never skills/. skills/ is generated from the root prompts by ./install.sh --build. After changing a prompt, rebuild and commit skills/ in the same commit.
Shared sections live in shared/. Some sections appear word for word in several prompts. Edit the file in shared/ and run python3 scripts/check-prompts.py --fix; editing one copy fails CI.
Contracts move with the prompts. If a prompt starts or stops reading or writing an artifact (PRD.md, FEATURES.md, RFCS.md, ...), update the contracts table in CONTRIBUTING.md in the same commit.
The example is real output. examples/url-shortener/after/ was produced by running the commands. Do not hand-edit it to match a prompt change; rerun the commands, or leave it and say it predates the change.
README changes reach the translations. The README.*.md translations (zh-CN, tr, ja, ko, es) follow README.md. Update them in the same change, keeping commands, code and file names in English, or tell the user they are now behind.
Versions move together. VERSION in install.sh, version in .claude-plugin/plugin.json, and the CHANGELOG entry.

Before committing, run:

python3 scripts/check-prompts.py
python3 -m unittest discover scripts
./install.sh --build && bash scripts/test-install.sh

When writing prompt text: keep it tool-agnostic and concrete, and say why a rule exists when the reason is not obvious — the existing prompts do this, and the reasons are what stop a later edit from "simplifying" the rule away.

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