create-prp — Claude Code command: researched, validated PRPs
Builds a Product Requirement Prompt by reading your docs and code, searching the web for examples, and writing an implementation guide.
npx claude-code-templates@latest --command project-management/create-prpWhat create-prp does
The command walks the agent through five stages: review the docs in ai_docs/, search the web for library references and working examples, study the base_template_v1 template and any PRPs already written, explore the codebase for patterns and integration points, then merge everything into one document. The output follows concept_library/cc_PRP_flow/PRPs/base_template_v1, and the command locates existing PRPs under concept_library/cc_PRP_flow/PRPs/ on its own.
What you get is more than a spec. A PRP carries file-level references, web resources, validation criteria and success metrics. The author sums up the idea as PRD plus curated codebase intelligence plus an agent runbook — the smallest packet an AI needs to ship production-ready code on the first pass.
Pass a feature description as the argument, or switch modes with --research, --template and --validate. Tool access is limited to Read, Write, Edit, WebSearch, Grep and Glob; there is no Bash, so nothing runs in your shell. The documentation and template stages need ai_docs/ and concept_library/cc_PRP_flow/ to exist — without them the research has nothing to stand on, so set up the structure first.
What it helps with
- pull feature requirements into one document before coding starts
- find library docs and implementation examples on the web
- map existing code into patterns and integration points
- define validation criteria and success metrics up front
- hand the agent enough context to ship working code on the first pass
Example prompts for the agent
- Create a PRP for CSV report export
- Run create-prp --research for the payments integration
- Use create-prp --validate on the PRP I already wrote
How to install create-prp
1claude-code-templates
npx claude-code-templates@latest --command project-management/create-prp2Manually: put the files in place
.claude/commands/create-prp.mdCall it with /create-prp. For all projects use ~/.claude/commands/
Manually: put the files in place
~/.codex/prompts/create-prp.mdCall it with /prompts:create-prp. Codex marks this format deprecated in favour of skills
Manually: put the files in place
.github/prompts/create-prp.prompt.mdManually: put the files in place
.gemini/commands/create-prp.tomlGemini stores commands as TOML: the prompt text goes into the prompt field
Questions about create-prp
How do I install create-prp in Claude Code?
Through the templates package: npx claude-code-templates@latest --command project-management/create-prp. The command file lands at .claude/commands/create-prp.md.
What does create-prp do?
It researches your documentation, the web and the codebase, then writes a Product Requirement Prompt from the base_template_v1 structure, complete with file references, validation criteria and success metrics.
Which tools can the command use?
Read, Write, Edit, WebSearch, Grep and Glob. Bash is not on the list, so it will not execute shell commands.
Source text
create-prp.md · MIT
--- allowed-tools: Read, Write, Edit, WebSearch, Grep, Glob argument-hint: "[feature-description] | --research | --template | --validate" description: Create comprehensive Product Requirement Prompt (PRP) with research and validation --- # Create Product Requirement Prompt Create comprehensive Product Requirement Prompt (PRP) following structured research process: **$ARGUMENTS** ## PRP Foundation - Base template: @concept_library/cc_PRP_flow/PRPs/base_template_v1 - PRP concept: @concept_library/cc_PRP_flow/README.md - Existing PRPs: !`find concept_library/cc_PRP_flow/PRPs/ -name "*.md" | head -5` - Documentation: @ai_docs/ directory analysis ## Task Develop comprehensive PRP through systematic research and structured documentation: **Research Process**: 1. **Documentation Review** - Analyze existing ai_docs/ and project documentation 2. **Web Research** - Gather implementation examples, library docs, and best practices 3. **Template Analysis** - Study base_template_v1 structure and existing PRPs 4. **Codebase Exploration** - Identify patterns, dependencies, and integration points 5. **Context Synthesis** - Compile comprehensive implementation context **PRP Development**: - Follow base_template_v1 structure exactly - Include specific file references and web resources - Provide curated codebase intelligence - Define clear validation criteria and success metrics - Create production-ready implementation guide **Remember**: PRP = PRD + curated codebase intelligence + agent/runbook—the minimum viable packet an AI needs to ship production-ready code on the first pass.
© davila7 · MIT · davila7/claude-code-templates
