Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Reverse-engineer a frontend codebase into a PRD. Usage: /code-to-prd [path]
.claude/skills/alirezarezvani-code-to-prd/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -16% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -41% | 0% |
Reverse-engineer a frontend codebase into a complete Product Requirements Document.
bash/code-to-prd # Analyze current project /code-to-prd ./src # Analyze specific directory /code-to-prd /path/to/project # Analyze external project
codebase_analyzer.py to detect framework, routes, APIs, enums, and project structureprd_scaffolder.py to create prd/ directory with README.md, per-page stubs, and appendix filesDetermine the project path (default: current directory). Run the frontend analyzer:
bashpython3 {skill_path}/scripts/codebase_analyzer.py {project_path} -o .code-to-prd-analysis.json
Display a summary of findings: framework, page count, API count, enum count.
Generate the PRD directory skeleton:
bashpython3 {skill_path}/scripts/prd_scaffolder.py .code-to-prd-analysis.json -o prd/
For each page in the inventory, follow the SKILL.md Phase 2 workflow:
prd/pages/ stubWork in batches of 3-5 pages for large projects (>15 pages). Ask the user to confirm after each batch.
Complete the appendix files:
prd/appendix/enum-dictionary.md — all enums and status codes foundprd/appendix/api-inventory.md — consolidated API referenceprd/appendix/page-relationships.md — navigation and data coupling mapClean up the temporary analysis file:
bashrm .code-to-prd-analysis.json
A prd/ directory containing:
README.md — system overview, module map, page inventorypages/*.md — one file per page with fields, interactions, APIsappendix/*.md — enum dictionary, API inventory, page relationshipsproduct-team/code-to-prd/skills/code-to-prd/SKILL.mdproduct-team/code-to-prd/skills/code-to-prd/scripts/codebase_analyzer.pyproduct-team/code-to-prd/skills/code-to-prd/scripts/prd_scaffolder.pyproduct-team/code-to-prd/skills/code-to-prd/references/prd-quality-checklist.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 3,627 | 6,267 | +73% | 1 | 1 | 0% | 188 | 1,048 | +457% | 0 | 0 | — |
case-02 | fail→fail | 11,523 | 5,520 | -52% | 1 | 1 | 0% | 2,092 | 910 | -57% | 0 | 0 | — |
case-03 | fail→fail | 4,777 | 6,212 | +30% | 1 | 1 | 0% | 162 | 945 | +483% | 0 | 0 | — |
case-04 | fail→pass | 6,009 | 3,325 | -45% | 1 | 1 | 0% | 1,087 | 1,359 | +25% | 0 | 0 | — |
case-05 | pass→pass | 13,957 | 2,215 | -84% | 1 | 1 | 0% | 2,517 | 1,027 | -59% | 0 | 0 | — |
case-06 | fail→pass | 9,169 | 2,761 | -70% | 1 | 1 | 0% | 1,472 | 923 | -37% | 0 | 0 | — |
case-07 | fail→pass | 12,892 | 4,952 | -62% | 1 | 1 | 0% | 1,913 | 1,605 | -16% | 0 | 0 | — |
case-08 | fail→fail | 9,205 | 2,116 | -77% | 1 | 1 | 0% | 1,622 | 975 | -40% | 0 | 0 | — |
case-09 | fail→fail | 8,893 | 1,460 | -84% | 1 | 1 | 0% | 1,671 | 901 | -46% | 0 | 0 | — |
case-10 | fail→pass | 10,884 | 2,274 | -79% | 1 | 1 | 0% | 1,963 | 1,013 | -48% | 0 | 0 | — |
case-11 | fail→pass | 15,210 | 2,891 | -81% | 1 | 1 | 0% | 2,078 | 1,225 | -41% | 0 | 0 | — |
case-16 | fail→pass | 5,417 | 2,821 | -48% | 1 | 1 | 0% | 947 | 1,127 | +19% | 0 | 0 | — |
case-12 | pass→pass | 3,757 | 2,318 | -38% | 1 | 1 | 0% | 505 | 938 | +86% | 0 | 0 | — |
case-13 | fail→pass | 8,676 | 1,845 | -79% | 1 | 1 | 0% | 1,499 | 957 | -36% | 0 | 0 | — |
case-14 | fail→pass | 11,317 | 10,222 | -10% | 1 | 1 | 0% | 2,082 | 2,133 | +2% | 0 | 0 | — |
case-15 | pass→pass | 12,901 | 8,716 | -32% | 1 | 1 | 0% | 2,250 | 2,193 | -3% | 0 | 0 | — |
case-17 | fail→fail | 7,465 | 1,861 | -75% | 1 | 1 | 0% | 1,184 | 947 | -20% | 0 | 0 | — |
case-18 | pass→pass | 9,035 | 3,376 | -63% | 1 | 1 | 0% | 1,379 | 996 | -28% | 0 | 0 | — |
case-19 | fail→pass | 21,883 | 1,854 | -92% | 1 | 1 | 0% | 1,424 | 1,035 | -27% | 0 | 0 | — |
case-20 | fail→fail | 27,914 | 2,661 | -90% | 1 | 1 | 0% | 5,742 | 1,043 | -82% | 0 | 0 | — |
case-21 | fail→fail | 5,652 | 6,248 | +11% | 1 | 1 | 0% | 1,004 | 1,765 | +76% | 0 | 0 | — |
case-22 | fail→fail | 9,056 | 14,166 | +56% | 1 | 1 | 0% | 2,009 | 2,905 | +45% | 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, and 19 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +41 percentage points is the difference between those two pass rates over the 19 comparable cases.
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.