All products

Write product decision documents with AI

Record a product choice, its alternatives and evidence, including what would reverse the decision and who needs to agree.

Write product decision documents with AI to preserve why one option won and what could change that choice. The decision-doc skill brings the question, deadline, alternatives and evidence into a record the team can revisit.

Keeping the status quo is a genuine option. Trade-offs can be explained in words instead of inventing precise-looking scores. An optional native reviewer challenges the document’s assumptions, but its simulated objections are not customer research. A proposed recommendation remains distinct from a decision that people have actually made.

What it does

  • Compare alternatives with the status quo
  • Explain selection criteria and trade-offs
  • Separate evidence from unresolved assumptions
  • Record reversibility and the cost of changing course
  • Identify required agreement and information roles
  • Preserve the reasoning for later review

Install and use

  1. Load the complete PMOS plugin in Claude Code, including its decision template and native reviewer. Provide the question, deadline, real alternatives and current evidence; the two-file leaf alone does not supply the reviewer.
  2. Ask: “Document a choice between export reports and better search. This month’s budget is fixed, and usage evidence is incomplete.”
  3. Review the criteria and reasons for rejecting alternatives. Keep missing data visible rather than filling gaps with invented scores, and specify what would change the recommendation.
  4. Request pressure testing if useful, then check the critique against the artifact. Separately record the human decision, necessary agreement and review date.

FAQ

Source

Made by Pawan Mehta. Original skill: decision-doc, published on GitHub under the MIT licence.

View on GitHub