ワンクリックで
pdd
// Transforms a rough idea into a detailed design document with implementation plan. Follows Prompt-Driven Development — iterative requirements clarification, research, design, and planning.
// Transforms a rough idea into a detailed design document with implementation plan. Follows Prompt-Driven Development — iterative requirements clarification, research, design, and planning.
Introspect, explain, and improve Ralph Orchestrator using its published llms.txt doc map. Use this skill whenever the user asks questions about Ralph's behavior, wants to understand how a Ralph internal works (event loop, hats, memories, tasks, backends, presets), debug an unfamiliar failure mode, or propose a code change to the ralph-orchestrator repo. The skill teaches the agent to discover authoritative answers from the live docs via llms.txt before guessing, and to scope improvements through the published architecture rather than the local checkout alone.
Run, monitor, resume, merge, and debug Ralph loops. Use this skill whenever the user asks to operate `ralph run` or `ralph loops`, inspect loop state, recover suspended loops, analyze diagnostics, or unblock merge queue issues.
Guides implementation of code tasks using test-driven development in an Explore, Plan, Code, Commit workflow. Acts as a Technical Implementation Partner and TDD Coach — following existing patterns, avoiding over-engineering, and producing idiomatic, modern code.
Generates structured .code-task.md files from descriptions or PDD implementation plans. Auto-detects input type, creates properly formatted tasks with Given-When-Then acceptance criteria.
Use when testing Ralph's hat collection presets, validating preset configurations, or auditing the preset library for bugs and UX issues.
Lists all code tasks in the repository with their status, dates, and metadata. Useful for getting an overview of pending work or finding specific tasks.
| name | pdd |
| description | Transforms a rough idea into a detailed design document with implementation plan. Follows Prompt-Driven Development — iterative requirements clarification, research, design, and planning. |
| type | anthropic-skill |
| version | 1.2 |
| metadata | {"internal":true} |
Transform a rough idea into a detailed design with an implementation plan. The process is iterative: clarify requirements, research, design, plan — moving between phases as needed.
These rules apply across ALL steps:
ralph run.specs/{task_name}/): Base directory for all artifacts. {task_name} is derived as kebab-case from the idea (e.g., "build a rate limiter" → rate-limiter). Aligns with Ralph's spec-driven pipeline.Constraints:
task_name from the rough idea as kebab-caseCreate the directory and initial files:
{project_dir}/rough-idea.md — the provided rough idea{project_dir}/requirements.md — Q&A record (initially empty){project_dir}/research/ — directory for research notesInform the user the structure feeds into Ralph's spec-driven presets.
Gate: You MUST NOT proceed to Step 2 until the user confirms the project structure is acceptable.
Ask the user their preferred starting point:
Gate: You MUST wait for the user to choose before proceeding. You MUST NOT automatically start requirements clarification or research without explicit user direction.
Guide the user through questions to refine the idea into a thorough specification.
Constraints:
Cover edge cases, user experience, technical constraints, and success criteria. Suggest options when the user is unsure.
Gate: You MUST NOT proceed to Research or Design until the user explicitly confirms requirements clarification is complete. You MUST offer the option to conduct research if questions arise that would benefit from additional information.
Conduct research on technologies, libraries, or existing code to inform the design.
Constraints:
{project_dir}/research/ as separate topic filesGate: You MUST NOT proceed to the Iteration Checkpoint until the user confirms research is sufficient. You MUST offer to return to requirements clarification if research uncovers new questions.
Summarize the current state of requirements and research, then ask the user:
Gate: You MUST NOT proceed to design without explicit user confirmation. You MUST support iterating between requirements and research as many times as needed.
Create {project_dir}/design.md as a standalone document with these sections:
Constraints:
Gate: You MUST NOT proceed to the implementation plan until the user explicitly approves the design. You MUST offer to return to requirements or research if gaps are identified during design review.
Create {project_dir}/plan.md — a numbered series of incremental implementation steps.
Guiding principle: Each step builds on previous steps, results in working demoable functionality, and follows TDD practices. No orphaned code — every step ends with integration. Core end-to-end functionality should be available as early as possible.
Constraints:
Gate: You MUST NOT proceed to the summary until the user reviews and approves the implementation plan.
Create {project_dir}/summary.md listing all artifacts, a brief overview, and suggested next steps. Present this summary in the conversation.
Ask: "Would you like me to create a PROMPT.md for Ralph to implement this autonomously?"
If yes, create a concise PROMPT.md (under 100 lines) with:
specs/{task_name}/Suggest the appropriate command:
ralph run --config presets/pdd-to-code-assist.ymlralph run -c ralph.yml -H builtin:code-assistIf the user declines, acknowledge and conclude the session.
Gate: You MUST NOT run ralph run or begin any implementation. This SOP ends here. Implementation is a separate step the user initiates themselves.
Input: "I want to build a template management feature for our internal tool — create, edit, share templates, generate documents with custom fields."
Output: A specs/template-management/ directory containing rough-idea.md, requirements.md, research/, design.md, plan.md, and summary.md — plus optionally a PROMPT.md for autonomous implementation.
Requirements stall: Suggest switching to a different aspect, provide examples, or pivot to research to unblock decisions.
Research limitations: Document what's missing, suggest alternatives with available information, ask user for additional context. Don't block progress.
Design complexity: Break into smaller components, focus on core functionality first, suggest phased implementation, return to requirements to re-prioritize.