Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create a Product Requirements Document using a comprehensive 8-section template covering problem, objectives, segments, value propositions, solution, and release planning. Use when writing a PRD, documenting product requirements, preparing a feature spec, or reviewing an existing PRD.
.claude/skills/phuryn-create-prd/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 14% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 97% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 40% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 44% | 0% |
You are an experienced product manager responsible for creating a comprehensive Product Requirements Document (PRD) for $ARGUMENTS. This document will serve as the authoritative specification for your product or feature, aligning stakeholders and guiding development.
A well-structured PRD clearly communicates the what, why, and how of your product initiative. This skill uses an 8-section template proven to communicate product vision effectively to engineers, designers, leadership, and stakeholders.
1. Summary (2-3 sentences)
2. Contacts
3. Background
4. Objective
5. Market Segment(s)
6. Value Proposition(s)
7. Solution
8. Release
PRD-[product-name].md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→fail | 34,761 | 20,761 | -40% | 1 | 1 | 0% | 6,160 | 4,349 | -29% | 0 | 0 | — |
case-01 | fail→fail | 29,573 | 23,331 | -21% | 1 | 1 | 0% | 4,613 | 4,766 | +3% | 0 | 0 | — |
case-03 | fail→pass | 24,839 | 32,426 | +31% | 1 | 1 | 0% | 4,130 | 4,726 | +14% | 0 | 0 | — |
case-04 | pass→pass | 9,498 | 11,944 | +26% | 1 | 1 | 0% | 1,900 | 3,397 | +79% | 0 | 0 | — |
case-05 | pass→fail | 15,379 | 22,483 | +46% | 1 | 1 | 0% | 2,564 | 4,773 | +86% | 0 | 0 | — |
case-06 | pass→fail | 26,913 | 25,156 | -7% | 1 | 1 | 0% | 5,779 | 5,630 | -3% | 0 | 0 | — |
case-07 | fail→fail | 14,007 | 17,047 | +22% | 1 | 1 | 0% | 2,483 | 3,851 | +55% | 0 | 0 | — |
case-08 | fail→pass | 20,081 | 39,195 | +95% | 1 | 1 | 0% | 3,293 | 4,092 | +24% | 0 | 0 | — |
case-09 | fail→fail | 31,865 | 39,863 | +25% | 1 | 1 | 0% | 5,716 | 6,957 | +22% | 0 | 0 | — |
case-10 | fail→pass | 14,375 | 16,040 | +12% | 1 | 1 | 0% | 1,897 | 3,740 | +97% | 0 | 0 | — |
case-11 | fail→fail | 13,871 | 17,925 | +29% | 1 | 1 | 0% | 2,463 | 3,970 | +61% | 0 | 0 | — |
case-12 | fail→fail | 8,860 | 13,387 | +51% | 1 | 1 | 0% | 1,471 | 3,072 | +109% | 0 | 0 | — |
case-13 | fail→fail | 16,314 | 20,985 | +29% | 1 | 1 | 0% | 2,262 | 3,455 | +53% | 0 | 0 | — |
case-14 | pass→pass | 46,422 | 20,870 | -55% | 1 | 1 | 0% | 3,884 | 4,527 | +17% | 0 | 0 | — |
case-15 | fail→fail | 15,110 | 26,570 | +76% | 1 | 1 | 0% | 2,263 | 5,859 | +159% | 0 | 0 | — |
case-16 | fail→fail | 19,162 | 16,507 | -14% | 1 | 1 | 0% | 2,304 | 3,837 | +67% | 0 | 0 | — |
case-17 | fail→pass | 24,288 | 23,429 | -4% | 1 | 1 | 0% | 3,756 | 5,256 | +40% | 0 | 0 | — |
case-18 | fail→fail | 17,992 | 22,290 | +24% | 1 | 1 | 0% | 3,271 | 3,900 | +19% | 0 | 0 | — |
case-19 | fail→pass | 19,144 | 19,503 | +2% | 1 | 1 | 0% | 3,033 | 4,355 | +44% | 0 | 0 | — |
case-20 | fail→fail | 9,582 | 14,473 | +51% | 1 | 1 | 0% | 1,684 | 3,431 | +104% | 0 | 0 | — |
case-21 | fail→fail | 12,979 | 14,616 | +13% | 1 | 1 | 0% | 1,875 | 3,034 | +62% | 0 | 0 | — |
case-22 | fail→pass | 17,381 | 22,795 | +31% | 1 | 1 | 0% | 2,982 | 4,157 | +39% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +18 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.