agentleFS
Sign inStart free
rpraharajrpraharaj/ai-sdlc/skills/write-requirements/SKILL.md

write-requirements

Skill0 starsChanged 16 days agoMITView on GitHub

Turn an approved PRD into testable functional requirements with traces and acceptance. Use when the user asks for requirements, a BRD, a functional spec, or /write-requirements. Do not write product code. Do not invent Must vs Should.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 3 are distinctive, and 3 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.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not implement. Do not invent users, SLAs, or priority.

    line 19
  2. Very specific:

    If there is no PRD (or the user cannot point at one), stop.

    line 33
  3. Distinctive:

    Must vs Should vs Later

    line 43
  4. Very specific:

    Read AGENTS.md if present.

    line 49
  5. Distinctive:

    Read the PRD and verify the authorized human's approval or waiver for this revision.

    line 50
  6. Very specific:

    No database-table design, framework choice, or folder layout (that is write-design).

    line 60

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

Write requirements

Capture the testable contract. This is not a PRD (write-prd), not a technical design (write-design), and not a ticket list (write-dev-tasks).

Do not implement. Do not invent users, SLAs, or priority.

Is this the right skill?

SignalDo this instead
Why / who / success still openwrite-prd
No PRD and the ask is a pitchwrite-opportunity
Current vs future processmap-process
Measurable quality targets onlywrite-nfr
How we will build itwrite-design
Ticketswrite-dev-tasks
One small already-clear changeshape-work

If there is no PRD (or the user cannot point at one), stop. Name write-prd. Do not write requirements from a conversation alone unless the user explicitly waives that and you record the waiver in the doc.

Required inputs

  • Path to the PRD (or the explicit waiver above)
  • AGENTS.md; context/glossary.md and context/systems-of-record.md if they exist
  • Human approval evidence for the PRD revision, or an explicit prerequisite waiver, following context/decision-rights.md. Missing, stale, or unassigned approval → stop even when a PRD file exists. An accepted label alone is insufficient.

Human decisions (do not invent)

  • Must vs Should vs Later
  • Who in the business is accountable for each Must
  • Legal / compliance wording

Procedure

  1. Read AGENTS.md if present. Read context/glossary.md and context/systems-of-record.md if present.
  2. Read the PRD and verify the authorized human's approval or waiver for this revision. If decision rights or the signer cannot be established, stop. List its goals and Must/Should if it has them. Do not add goals the PRD does not have.
  3. Find the home: first existing docs/requirements/, a path AGENTS.md names, or offer to create docs/requirements/.
  4. Name: docs/requirements/YYYY-MM-DD-<slug>.md.
  5. Open references/requirements-template.md. Copy every heading. Write n/a when unused.
  6. Each requirement:
    • id R-…
    • traces to a PRD goal (or a named PRD section)
    • acceptance that a stranger can check (command, UI observation, or API response)
    • priority only if the user or the PRD already set it; otherwise unassigned and stop for Must vs Should
  7. Systems named must appear in context/systems-of-record.md or be marked not in systems-of-record.
  8. No database-table design, framework choice, or folder layout (that is write-design). The requirements template's evidence tables are required. No secrets.

Output

  • Path: docs/requirements/YYYY-MM-DD-<slug>.md
  • Template: references/requirements-template.md

Quality bar

  • PRD path is named at the top, or an explicit waiver is named
  • The PRD revision has authorized human approval evidence, or a recorded prerequisite waiver; the agent did not grant approval
  • Every Must traces to a PRD goal
  • Every Must has a yes/no acceptance check
  • At least one acceptance check would fail today
  • Priority is not invented (unassigned or taken from the PRD/user)
  • Unused headings are n/a

Done

  1. Path
  2. Count of Must / Should / unassigned
  3. Next skill: map-process, write-nfr, write-architecture (new boundaries), or write-design

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