共 118 个技能可用
The canonical knowledge base for designing Agent Skills. Contains distilled Agent Skills specifications (naming conventions, frontmatter rules, directory structure, progressive disclosure), official best practices, description optimization techniques, script safety guidelines, and evaluation frameworks (TRACE). Use when designing a new skill, reviewing an existing skill for compliance, needing to know what makes a good skill, wondering about frontmatter rules or naming conventions, or when the user mentions "设计技能", "创建技能", "技能规范", "技能命名规则", "SKILL.md 怎么写", "frontmatter 规范", "技能最佳实践", "技能目录结构", "渐进式披露", "Agent Skills 规范".
创建新技能、修改和改进现有技能，并衡量技能表现。当用户想从零创建技能、编辑或优化现有技能、运行评测来测试技能时使用。
Manages the installation and discovery of AI skills from the PartMe marketplace.
Guide for creating Standard Operating Procedures and SOPs and composite workflows by combining atomic Agent Skills. Use when users want to design, structure, or implement multi-step workflows that orchestrate multiple skills for complex business goals. Transforms loose tool collections into reliable, repeatable business processes.
对任意 Agent Skill 做 TRACE 五维度评测（T/R/A/C/E），输出 Markdown/HTML 报告与雷达图；当用户要求“TRACE 评测/打分/生成 TRACE 报告/五维度评估”或需要改进建议时使用。
将每一个实现决策都建立在官方文档之上。当你想要权威的、有出处引用的、不含过时模式的代码时使用。
在编码之前先创建规格。当开始一个新项目、新功能或重大变更，且尚不存在规格说明时使用。
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
Generate feature specifications by analyzing existing source code.