agentleFS
Sign inStart free
sson0-ersson0-er/claudex/skills/define-requirements/SKILL.md

define-requirements

Skill0 starsChanged 53 days agoMITView on GitHub

Work with the user, one question at a time, to complete the requirements document (docs/claudex/<slug>/01-requirements.md). Run by the orchestrator (parent). Stays out of design and implementation.

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 2 are distinctive, and 2 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.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run claudex-init $slug and open docs/claudex/$slug/01-requirements.md.

    line 16
  2. Distinctive:

    Ask one question per message.

    line 18
  3. Very specific:

    After each answer, update the document immediately and remove the (to confirm) marker.

    line 23
  4. Distinctive:

    When Open questions is empty and at least one acceptance criterion is observable, show the whole document and ask the user to approve it.

    line 24

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill 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
SKILL.md

Requirements definition: $slug

You are the facilitator. The user decides the requirements; you find gaps, ask, and write down what was agreed. Talk to the user in the user's language. The requirements document itself is written in English.

Procedure

  1. Run claudex-init $slug and open docs/claudex/$slug/01-requirements.md. If it already has content, continue from there.
  2. From the user's initial request, fill every section of the template you can (background and goal, scope, functional and non-functional requirements, acceptance criteria). Mark anything you inferred with (to confirm).
  3. Ask one question per message. Use AskUserQuestion when the answer is a choice, with 3 or 4 options. Priority order:
    1. Unclear purpose (whose problem, which problem)
    2. Scope boundaries (items where in/out could go either way)
    3. Acceptance criteria (what the user will look at to call it done)
    4. Non-functional requirements (performance, security, compatibility)
  4. After each answer, update the document immediately and remove the (to confirm) marker.
  5. When Open questions is empty and at least one acceptance criterion is observable, show the whole document and ask the user to approve it. This is gate 1.
  6. On approval, write Status: approved (<yyyy-mm-dd>) at the top of the document and report "gate 1 passed" to the orchestrator.

Do not

  • Discuss design or technology choices. If the user starts, say it belongs to the design phase and return to requirements.
  • Bundle several questions into one message.
  • Fill gaps by guessing and treating the guess as settled. Ask.

Download now (ZIP)

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 2

    1. sson0-er/orchestraude★ 096% same
    2. sson0-er/sonnede★ 096% same

    Wrote their own version · 0

    None found.

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository2 other files it gives its agents.

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.

Check your setup, free