一键导入
ralph-validate-prd
PRD quality checklist that validates feature completeness, test coverage, sizing, dependency ordering, and acceptance criteria quality.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
PRD quality checklist that validates feature completeness, test coverage, sizing, dependency ordering, and acceptance criteria quality.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Atomically claim a feature for a teammate. Uses file locking to prevent two teammates from claiming the same feature. Teams variant only.
Get a compact summary of project status and feature counts from feature_list.json.
Select the next feature to implement from feature_list.json based on priority rules (in_progress first, then first pending).
Increment the attempts counter for a feature after a failed implementation attempt. Optionally sets the last_error message.
Interactive assistant that guides users through creating high-quality PRDs for Ralph-RLM Framework. Asks structured questions about project type, requirements, testing needs, and dependency ordering.
Update the status of a feature in feature_list.json (pending, in_progress, complete, or blocked) and recalculate stats.
| name | ralph-validate-prd |
| description | PRD quality checklist that validates feature completeness, test coverage, sizing, dependency ordering, and acceptance criteria quality. |
| user-invocable | false |
Validate a PRD for completeness and quality before running Ralph initialization.
Pre-flight quality check for prd.md to catch issues before the Initializer decomposes it into features. Finding problems here saves iteration cycles later.
dotnet test, npm test, pytest)## headings for section discovery--- dividers between sections| Issue | Example | Fix |
|---|---|---|
| Too vague | "System should work well" | "Login API must return 200 within 500ms" |
| Too large | "Build the dashboard" | Split into individual components |
| Missing errors | Only happy path described | Add "When X fails, the system must Y" |
| No tests | No mention of testing | Add unit/integration/E2E test requirements |
| Wrong order | Controllers before models | Reorder: models -> services -> controllers |
| Missing NFRs | No performance targets | Add specific numbers for latency, throughput |