All products

Write a product requirements document with AI

Turn a feature idea into a product brief with a problem statement, scope, acceptance criteria, success measures and decisions the team still needs to make.

To write a product requirements document with AI, give write-spec a user problem and a situation in which it occurs. A teacher who cannot find the right worksheet during a lesson is a sharper starting point than a request to improve search. The agent can use that example to define the outcome and suggest acceptance criteria.

The resulting document makes scope decisions visible. A first version might support searching by lesson name while leaving document history for later. If the team has not decided who can see archived worksheets, that belongs among the open questions. Keeping such decisions visible makes it easier to review the plan with design and engineering.

What it does

  • Describe the user and the problem in context.
  • Separate essential requirements from later ideas.
  • Write user stories with observable outcomes.
  • Define acceptance criteria for the proposed behaviour.
  • Choose success measures and record unresolved questions.

Install and use

  1. Install the skill and provide your product context or research notes. Working from supplied material requires no extra account.
  2. Ask “Draft a requirements document for lesson-name search. Keep file history outside the first release and list decisions we still need.”
  3. Review the scope and criteria with the people who will build and evaluate the feature.

FAQ

Source

Made by Anthropic. Original skill: write-spec, published on GitHub under the Apache-2.0 licence.

View on GitHub