kiro-validate-design
Interactive technical design quality review and validation. Use when reviewing design before implementation.
SKILL.mdskills/kiro/phases/validate-design/SKILL.md
GitHub移植说明|本文件是
kiroskill 的阶段规范(源自kiro-validate-designskill),由 kiro 总控按需读取并就地执行。 文中出现的kiro-xxx//kiro-xxxskill 引用,一律按kiro/SKILL.md的「阶段解析规则」处理: 宿主已注册同名 skill →Skill(kiro-xxx);否则读phases/xxx/SKILL.md就地执行。 相对路径(rules/、templates/)一律相对本文件所在目录phases/validate-design/。
kiro-validate-design Skill
Role
You are a specialized skill for conducting interactive quality review of technical design to ensure readiness for implementation.
Core Mission
- Mission: Conduct interactive quality review of technical design to ensure readiness for implementation
- Success Criteria:
- Critical issues identified (maximum 3 most important concerns)
- Balanced assessment with strengths recognized
- Clear GO/NO-GO decision with rationale
- Actionable feedback for improvements if needed
Execution Steps
Step 1: Gather Context
If steering/spec context is already available from conversation, skip redundant file reads. Otherwise, load all necessary context:
- Read
.kiro/specs/{feature}/spec.jsonfor language and metadata - Read
.kiro/specs/{feature}/requirements.mdfor requirements - Read
.kiro/specs/{feature}/design.mdfor design document - Core steering context:
product.md,tech.md,structure.md - Additional steering files only when directly relevant to architecture boundaries, integrations, runtime prerequisites, domain rules, security/performance constraints, or team conventions that affect implementation readiness
- Relevant local agent skills or playbooks only when they clearly match the feature's host environment or use case and provide review-relevant context
Parallel Research (concurrent subagent dispatch)
Follow ../../rules/parallel-dispatch.md. Dispatch as concurrent subagents in a single message (subagent_type: "Explore"):
- Context & rules loading: Spec documents, core steering, feature-relevant extra steering, relevant local agent playbooks, and
rules/design-review.mdfor review criteria - Codebase pattern survey: Existing architecture patterns, naming conventions, component structure → the reference the design will be judged against
- Feasibility spot-check: For each external contract / API / library the design depends on, verify it actually exists and works as the design assumes (
general-purposewhen this needs commands or web access) → the single highest-value dimension, since designs fail most often on assumptions that were never checked
After all return, synthesize in the parent context. The GO/NO-GO call and the "3 most important concerns" cut are the parent's job — a subagent only sees its own dimension and cannot rank across them.
Step 2: Execute Design Review
- Reference conversation history: leverage prior requirements discussion and user's stated design intent
- Follow design-review.md process: Analysis → Critical Issues → Strengths → GO/NO-GO
- Limit to 3 most important concerns
- Engage interactively with user — ask clarifying questions, propose alternatives
- Use language specified in spec.json for output
Step 3: Decision and Next Steps
- Clear GO/NO-GO decision with rationale
- Provide specific actionable next steps (see Next Phase below)
Important Constraints
- Quality assurance, not perfection seeking: Accept acceptable risk
- Critical focus only: Maximum 3 issues, only those significantly impacting success
- Conversation-aware: Leverage discussion history for requirements context and user intent
- Interactive approach: Engage in dialogue, ask clarifying questions, propose alternatives
- Balanced assessment: Recognize both strengths and weaknesses
- Actionable feedback: All suggestions must be implementable
- Context Discipline: Start with core steering and expand only with review-relevant steering or use-case-aligned local agent skills/playbooks
Tool Guidance
- Read first: Load spec, core steering, relevant local playbooks/agent skills, and rules before review
- Grep if needed: Search codebase for pattern validation or integration checks
- Interactive: Engage with user throughout the review process
Output Description
Provide output in the language specified in spec.json with:
- Review Summary: Brief overview (2-3 sentences) of design quality and readiness
- Critical Issues: Maximum 3, following design-review.md format
- Design Strengths: 1-2 positive aspects
- Final Assessment: GO/NO-GO decision with rationale and next steps
Format Requirements:
- Use Markdown headings for clarity
- Follow design-review.md output format
- Keep summary concise
Safety & Fallback
Error Scenarios
- Missing Design: If design.md doesn't exist, stop with message: "Run
/kiro-spec-design {feature}first to generate design document" - Design Not Generated: If design phase not marked as generated in spec.json, warn but proceed with review
- Empty Steering Directory: Warn user that project context is missing and may affect review quality
- Language Undefined: Default to English (
en) if spec.json doesn't specify language
Next Phase: Task Generation
If Design Passes Validation (GO Decision):
- Apply any suggested improvements if agreed
- Run
/kiro-spec-tasks {feature}to generate implementation tasks - Or
/kiro-spec-tasks {feature} -yto auto-approve and proceed directly
If Design Needs Revision (NO-GO Decision):
- Address critical issues identified in review
- Re-run
/kiro-spec-design {feature}with improvements - Re-validate with
/kiro-validate-design {feature}
Note: Design validation is recommended but optional. Quality review helps catch issues early.