Back to Details

skill-architect

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Skill Architect

Skill 编排器

Purpose

用途

Meta-orchestrator for the complete Claude Code skill lifecycle. Manages skill creation, review, optimization, and quality assurance through coordinated subagent workflows.
用于Claude Code Skill完整生命周期的元编排器。通过协调子Agent工作流来管理Skill的创建、评审、优化和质量保证。

When to Use

使用场景

  • User asks to "manage skills" or "oversee skill development"
  • Complete skill lifecycle management
  • Quality assurance for skill libraries
  • Team skill development workflows
  • Comprehensive skill audits
  • 用户提出"管理Skills"或"监督Skill开发"需求时
  • 完整Skill生命周期管理
  • Skill库的质量保证
  • 团队Skill开发工作流
  • 全面的Skill审计

Available Subagents

可用子Agent

This orchestrator coordinates three specialized skills:
  1. @skill-generator - Creates new skills from descriptions
  2. @skill-reviewer - Reviews skills for quality and best practices
  3. @skill-optimizer - Refactors skills using PDA
该编排器协调三个专业Skill：
  1. @skill-generator - 根据描述创建新的Skill
  2. @skill-reviewer - 评审Skill的质量和最佳实践
  3. @skill-optimizer - 使用PDA重构Skill

Workflows

工作流

Workflow 1: Create and Review

工作流1：创建与评审

Use when: Creating a new skill with quality assurance
1. @skill-generator
   - Generate SKILL.md from user description
   - Create directory structure
   - Apply best practices

2. @skill-reviewer
   - Review generated skill
   - Identify any issues
   - Provide quality score

3. Report and Iterate (if needed)
   - Present skill to user
   - Address review feedback
   - Finalize skill
Trigger phrases:
  • "Create a skill for X and review it"
  • "Generate and check a new skill"
  • "Make a skill for X with quality check"
适用场景： 创建带质量保证的新Skill
1. @skill-generator
   - 根据用户描述生成SKILL.md
   - 创建目录结构
   - 应用最佳实践

2. @skill-reviewer
   - 评审生成的Skill
   - 识别问题
   - 提供质量评分

3. 报告与迭代（如有需要）
   - 向用户展示Skill
   - 处理评审反馈
   - 最终确定Skill
触发短语：
  • "为X创建一个Skill并进行评审"
  • "生成并检查新Skill"
  • "为X创建带质量检查的Skill"

Workflow 2: Audit and Optimize

工作流2：审计与优化

Use when: Improving existing skills
1. @skill-reviewer
   - Review existing skill
   - Identify improvement opportunities
   - Calculate potential savings

2. @skill-optimizer
   - Apply recommended improvements
   - Implement PDA refactoring
   - Optimize token usage

3. Validation
   - Compare before/after
   - Verify functionality preserved
   - Report improvements
Trigger phrases:
  • "Audit this skill and optimize it"
  • "Review and improve this skill"
  • "Check and fix this skill"
适用场景： 改进现有Skill
1. @skill-reviewer
   - 评审现有Skill
   - 识别改进机会
   - 计算潜在成本节约

2. @skill-optimizer
   - 应用推荐的改进方案
   - 实施PDA重构
   - 优化Token使用

3. 验证
   - 对比优化前后的差异
   - 确认功能保留
   - 报告改进成果
触发短语：
  • "审计并优化这个Skill"
  • "评审并改进这个Skill"
  • "检查并修复这个Skill"

Workflow 3: Complete Lifecycle

工作流3：完整生命周期

Use when: End-to-end skill development
1. Requirements Gathering
   - Understand skill purpose
   - Identify use cases
   - Determine scope

2. @skill-generator
   - Create initial skill
   - Structure content
   - Add examples

3. @skill-reviewer
   - Quality assessment
   - Best practices check
   - Improvement recommendations

4. @skill-optimizer (if needed)
   - Apply optimizations
   - Implement PDA
   - Reduce token usage

5. Final Review
   - Validate improvements
   - Confirm quality standards
   - Deliver production-ready skill
Trigger phrases:
  • "Create a production-ready skill for X"
  • "Develop a complete skill from scratch"
  • "Build and validate a skill"
适用场景： 端到端Skill开发
1. 需求收集
   - 明确Skill用途
   - 识别使用场景
   - 确定范围

2. @skill-generator
   - 创建初始Skill
   - 构建内容结构
   - 添加示例

3. @skill-reviewer
   - 质量评估
   - 最佳实践检查
   - 提出改进建议

4. @skill-optimizer（如有需要）
   - 应用优化方案
   - 实施PDA
   - 减少Token使用

5. 最终评审
   - 验证改进效果
   - 确认符合质量标准
   - 交付生产可用的Skill
触发短语：
  • "为X创建生产可用的Skill"
  • "从零开始开发完整的Skill"
  • "构建并验证一个Skill"

Workflow 4: Skill Library Audit

工作流4：Skill库审计

Use when: Reviewing multiple skills
1. Discover Skills
   - Find all skills in .claude/skills/
   - List each skill with metadata

2. @skill-reviewer (parallel)
   - Review each skill
   - Generate quality scores
   - Identify common issues

3. Aggregate Report
   - Overall quality assessment
   - Prioritized improvements
   - Token savings opportunities

4. Recommendations
   - Which skills need optimization
   - Which skills are excellent
   - Best practices to share
Trigger phrases:
  • "Audit all my skills"
  • "Review my skill library"
  • "Check all skills for quality"
适用场景： 评审多个Skill
1. 发现Skill
   - 查找.claude/skills/下的所有Skill
   - 列出每个Skill及其元数据

2. @skill-reviewer（并行处理）
   - 评审每个Skill
   - 生成质量评分
   - 识别常见问题

3. 汇总报告
   - 整体质量评估
   - 按优先级排序的改进项
   - Token节约机会

4. 建议
   - 哪些Skill需要优化
   - 哪些Skill表现优秀
   - 可共享的最佳实践
触发短语：
  • "审计我所有的Skill"
  • "评审我的Skill库"
  • "检查所有Skill的质量"

Output Format

输出格式

For Creation Workflows

创建工作流输出格式

markdown
undefined
markdown
undefined

Skill Creation Complete

Skill 创建完成

Skill Overview

Skill 概述

Name: [skill-name] Location: .claude/skills/[skill-name]/ Size: [X] KB
名称: [skill-name] 位置: .claude/skills/[skill-name]/ 大小: [X] KB

Quality Assessment

质量评估

Score: [X/10] PDA Compliance: Yes/No Status: Ready for use / Needs improvements
评分: [X/10] PDA合规性: 是/否 状态: 可直接使用 / 需要改进

What Was Created

创建内容

  1. SKILL.md - Main orchestrator
  2. reference/ - Detailed documentation
  3. scripts/ - Utility scripts (if applicable)
  1. SKILL.md - 主编排器
  2. reference/ - 详细文档
  3. scripts/ - 实用脚本（如适用）

Usage

使用方法

Invoke with: [example invocation]
调用方式: [示例调用]

Next Steps

后续步骤

  • Test the skill
  • Adjust if needed
  • Add to version control
undefined
  • 测试该Skill
  • 按需调整
  • 添加到版本控制
undefined

For Optimization Workflows

优化工作流输出格式

markdown
undefined
markdown
undefined

Skill Optimization Complete

Skill 优化完成

Summary

摘要

Skill: [skill-name] Original Size: [X] KB Optimized Size: [Y] KB Token Savings: [Z]%
Skill: [skill-name] 原大小: [X] KB 优化后大小: [Y] KB Token节约率: [Z]%

Improvements Made

已实施的改进

Structure

结构

  • [Changes]
  • [变更内容]

Content

内容

  • [Changes]
  • [变更内容]

Token Efficiency

Token效率

  • [Changes]
  • [变更内容]

Quality Improvements

质量提升

Before: Score [X]/10 After: Score [Y]/10
优化前: 评分 [X]/10 优化后: 评分 [Y]/10

Cost Impact

成本影响

At 100 requests/day:
  • Before: $[cost]/year
  • After: $[new cost]/year
  • Saved: $[savings]/year
按每日100次请求计算:
  • 优化前: $[cost]/年
  • 优化后: $[new cost]/年
  • 节约: $[savings]/年

Validation

验证结果

✅ Functionality preserved ✅ All links work ✅ Performance improved
undefined
✅ 功能保留完整 ✅ 所有链接可用 ✅ 性能提升
undefined

For Audit Workflows

审计工作流输出格式

markdown
undefined
markdown
undefined

Skill Library Audit Report

Skill 库审计报告

Overview

概述

Total Skills: [N] Average Score: [X]/10 Skills Needing Work: [N]
总Skill数量: [N] 平均评分: [X]/10 需要改进的Skill数量: [N]

Skills by Quality

Skill 质量分级

Excellent (9-10)

优秀 (9-10)

  • [skill-1]
  • [skill-2]
  • [skill-1]
  • [skill-2]

Good (7-8)

良好 (7-8)

  • [skill-3]
  • [skill-4]
  • [skill-3]
  • [skill-4]

Needs Improvement (5-6)

需要改进 (5-6)

  • [skill-5]
  • [skill-6]
  • [skill-5]
  • [skill-6]

Poor (1-4)

较差 (1-4)

  • [skill-7]
  • [skill-8]
  • [skill-7]
  • [skill-8]

Common Issues Found

发现的常见问题

  1. [Issue] - [Count] skills affected
  2. [Issue] - [Count] skills affected
  1. [问题] - 影响[Count]个Skill
  2. [问题] - 影响[Count]个Skill

Prioritized Recommendations

优先级建议

P1 (Critical)

P1（关键）

  • [Recommendations]
  • [建议内容]

P2 (Important)

P2（重要）

  • [Recommendations]
  • [建议内容]

P3 (Nice-to-Have)

P3（可选）

  • [Recommendations]
  • [建议内容]

Token Optimization Opportunity

Token优化机会

Current Usage: [X] MB/day Potential Savings: [Y] MB/day ([Z]%) Annual Cost Savings: $[amount]
当前使用量: [X] MB/天 潜在节约量: [Y] MB/天 ([Z]%) 年度成本节约: $[amount]

Next Steps

后续步骤

  1. Address P1 issues
  2. Optimize skills with low scores
  3. Apply best practices to all skills
undefined
  1. 处理P1级问题
  2. 优化低评分Skill
  3. 为所有Skill应用最佳实践
undefined

Quality Gates

质量关卡

Each workflow includes quality checks:
每个工作流都包含质量检查：

Creation Quality Gates

创建质量关卡

  • YAML frontmatter valid
  • Description specific and actionable
  • Clear instructions with examples
  • Appropriate use of PDA
  • No reserved words in name
  • File paths use forward slashes
  • YAML前置元数据有效
  • 描述具体且可执行
  • 包含清晰的示例说明
  • 合理使用PDA
  • 名称中无保留字
  • 文件路径使用正斜杠

Optimization Quality Gates

优化质量关卡

  • Token savings achieved
  • Functionality preserved
  • Links verified working
  • Structure improved
  • User experience enhanced
  • 实现Token节约
  • 功能保留完整
  • 链接验证可用
  • 结构得到改进
  • 用户体验提升

Review Quality Gates

评审质量关卡

  • All dimensions evaluated
  • Severity classification applied
  • Recommendations actionable
  • Token savings calculated
  • Report comprehensive
  • 评估所有维度
  • 应用严重程度分类
  • 建议可执行
  • 计算Token节约量
  • 报告内容全面

Best Practices Applied

应用的最佳实践

This orchestrator ensures:
  1. Progressive Disclosure
    • Skills use PDA when appropriate
    • Token-efficient design
    • On-demand loading
  2. Quality Standards
    • Consistent structure
    • Clear documentation
    • Actionable examples
  3. Best Practices
    • Gerund naming
    • Third-person descriptions
    • Forward slash paths
    • Single responsibility
  4. Token Optimization
    • Minimal SKILL.md size
    • Reference files for details
    • Scripts for mechanical work
该编排器确保：
  1. 渐进式披露（Progressive Disclosure）
    • Skill在合适场景下使用PDA
    • 高效Token设计
    • 按需加载
  2. 质量标准
    • 一致的结构
    • 清晰的文档
    • 可执行的示例
  3. 最佳实践
    • 动名词命名
    • 第三人称描述
    • 正斜杠路径
    • 单一职责
  4. Token优化
    • 最小化SKILL.md大小
    • 使用参考文件存储详细内容
    • 脚本处理机械性工作

Integration

集成

This orchestrator works with:
  • skill-generator - For skill creation
  • skill-reviewer - For quality assessment
  • skill-optimizer - For improvement
该编排器可与以下组件配合使用：
  • skill-generator - 用于Skill创建
  • skill-reviewer - 用于质量评估
  • skill-optimizer - 用于Skill改进

Advanced Features

高级功能

Batch Operations

批量操作

Process multiple skills:
> Audit and optimize all skills in .claude/skills/
处理多个Skill：
> 审计并优化.claude/skills/下的所有Skill

Runs @skill-reviewer on each

对每个Skill运行@skill-reviewer

Runs @skill-optimizer on those needing improvement

对需要改进的Skill运行@skill-optimizer

Generates aggregate report

生成汇总报告

undefined
undefined

Skill Templates

Skill模板

Create from templates:
> Create a generator skill following the generator template
基于模板创建：
> 按照生成器模板创建一个生成器Skill

Uses predefined pattern

使用预定义模式

Applies best practices automatically

自动应用最佳实践

Delivers consistent structure

交付一致的结构

undefined
undefined

Continuous Improvement

持续改进

Ongoing quality management:
> Set up skill quality monitoring
持续质量管理：
> 设置Skill质量监控

Establishes quality baseline

建立质量基线

Recommends regular reviews

建议定期评审

Tracks improvements over time

跟踪长期改进效果

undefined
undefined

Examples

示例

Example 1: Create Complete Skill

示例1：创建完整Skill

User: "Create a production-ready skill for processing JSON files with validation and transformation"
Workflow:
1. @skill-generator
   Creates: json-processor/
   ├── SKILL.md (4KB orchestrator)
   ├── reference/
   │   ├── json-schema.md
   │   ├── transformations.md
   │   └── validation.md
   └── scripts/
       ├── validate.py
       └── transform.py

2. @skill-reviewer
   Score: 9/10
   PDA Compliance: Yes
   Status: Ready for use

3. Report
   Skill created and validated
   All quality gates passed
   Ready for production use
用户: "创建一个用于处理JSON文件并包含验证和转换功能的生产可用Skill"
工作流:
1. @skill-generator
   创建: json-processor/
   ├── SKILL.md (4KB 主编排器)
   ├── reference/
   │   ├── json-schema.md
   │   ├── transformations.md
   │   └── validation.md
   └── scripts/
       ├── validate.py
       └── transform.py

2. @skill-reviewer
   评分: 9/10
   PDA合规性: 是
   状态: 可直接使用

3. 报告
   Skill已创建并验证
   所有质量关卡通过
   可投入生产使用

Example 2: Optimize Poor Skill

示例2：优化低质量Skill

User: "This 50KB skill is slow and expensive. Fix it."
Workflow:
1. @skill-reviewer
   Score: 3/10
   Issues: Monolithic, no PDA, token-inefficient
   Recommendation: Major refactor

2. @skill-optimizer
   Before: 50KB SKILL.md
   After: 3KB SKILL.md + reference/ files
   Savings: 84%

3. Validation
   Functionality: Preserved
   Performance: 5x faster
   Cost: $12/day → $2/day
用户: "这个50KB的Skill速度慢且成本高。修复它。"
工作流:
1. @skill-reviewer
   评分: 3/10
   问题: 单体结构、未使用PDA、Token效率低
   建议: 重大重构

2. @skill-optimizer
   优化前: 50KB SKILL.md
   优化后: 3KB SKILL.md + reference/文件
   节约率: 84%

3. 验证
   功能: 完整保留
   性能: 提升5倍
   成本: $12/天 → $2/天

Example 3: Library Audit

示例3：库审计

User: "Review all my skills and tell me what needs improvement"
Workflow:
1. Discovery
   Found 12 skills

2. @skill-reviewer (parallel)
   3 Excellent (9-10)
   5 Good (7-8)
   3 Fair (5-6)
   1 Poor (2)

3. @skill-optimizer (for 4 low-scoring skills)
   Average improvement: 73% token reduction

4. Report
   Overall quality improved from 6.5 to 8.2
   Projected annual savings: $1,200
用户: "评审我所有的Skill并告诉我哪些需要改进"
工作流:
1. 发现
   找到12个Skill

2. @skill-reviewer（并行处理）
   3个优秀 (9-10)
   5个良好 (7-8)
   3个一般 (5-6)
   1个较差 (2)

3. @skill-optimizer（针对4个低评分Skill）
   平均Token减少率: 73%

4. 报告
   整体质量从6.5提升至8.2
   预计年度成本节约: $1,200

See Also

另请参阅

  • SKILL_GENERATOR.md - Create skills
  • SKILL_REVIEWER.md - Review skills
  • SKILL_OPTIMIZER.md - Optimize skills
  • CLAUDE_SKILLS_ARCHITECTURE.md - Reference documentation
  • docs/README.md - All documentation
  • SKILL_GENERATOR.md - 创建Skill
  • SKILL_REVIEWER.md - 评审Skill
  • SKILL_OPTIMIZER.md - 优化Skill
  • CLAUDE_SKILLS_ARCHITECTURE.md - 参考文档
  • docs/README.md - 全部文档

Sources

参考来源

Based on:
  • CLAUDE_SKILLS_ARCHITECTURE.md
  • AGENTS_WORKFLOWS.md
  • Progressive Disclosure Architecture principles
基于以下内容:
  • CLAUDE_SKILLS_ARCHITECTURE.md
  • AGENTS_WORKFLOWS.md
  • 渐进式披露架构（Progressive Disclosure Architecture）原则