Back to Details

skill-creator

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Skill Authoring Procedure

技能创作流程

Create professional-grade skills with lean context, deterministic structure, and validation.
打造具备精简上下文、确定性结构及验证机制的专业级技能。

When to Use This Skill

何时使用本技能

  • User wants to create a new skill directory
  • User wants to improve a skill's discoverability or metadata
  • User wants to split large instructions into references or scripts
  • User wants to add or update skillgrade validation
  • 用户希望创建新的技能目录
  • 用户希望提升技能的可发现性或优化元数据
  • 用户希望将冗长的指令拆分至参考资料或脚本中
  • 用户希望添加或更新skillgrade验证机制

Procedures

操作步骤

Step 1: Validate the Skill Metadata

步骤1：验证技能元数据

Check that the frontmatter uses a unique lowercase name, a specific description, and clear negative triggers.
Keep the description short enough to fit within the agent router's metadata budget.
检查前置元数据是否使用了唯一的小写名称、具体的描述以及清晰的负面触发条件。
确保描述足够简短，以适配Agent路由的元数据预算。

Step 2: Keep the Main Skill Lean

步骤2：保持主技能精简

Write the main 
SKILL.md
as a high-level workflow.
Move dense rules, large schemas, and reusable templates into 
references/
or 
assets/
.
Use 
scripts/
only for fragile or repetitive logic that should not be re-authored by the agent.
将主文件
SKILL.md
编写为高层级工作流。
将密集的规则、大型模式及可复用模板移至
references/
或
assets/
目录下。
仅将脆弱或重复性逻辑放入
scripts/
目录，避免Agent重复编写此类逻辑。

Step 3: Use Progressive Disclosure

步骤3：采用渐进式披露机制

Command the agent to read supporting files only when the current step needs them.
Prefer one-level-deep support files with explicit relative paths.
When a skill depends on repository state that will differ at invocation time, prefer a guarded inline shell snapshot such as 
!`node -e "..."`
instead of baking the current state into prose.
Only do this when the command materially improves the invoked prompt, and keep the output short and deterministic.
If a skill uses inline commands, declare the minimum required 
allowed-tools
entries in frontmatter and keep them read-only.
仅当当前步骤需要时，才命令Agent读取支持文件。
优先使用一级深度的支持文件，并指定明确的相对路径。
当技能依赖于调用时会变化的仓库状态时，优先使用受保护的内联Shell快照，例如
!`node -e "..."`
，而非将当前状态硬编码到文本中。
仅当该命令能切实提升调用提示的质量时才这么做，且需保证输出简短且具有确定性。
如果技能使用了内联命令，需在前置元数据中声明所需的最低权限
allowed-tools
条目，并保持其只读属性。

Step 4: Add Validation

步骤4：添加验证机制

Create a 
skillgrade
eval when the skill needs regression testing.
Use a deterministic grader for structural checks and an LLM rubric only when qualitative judgment is necessary.
当技能需要回归测试时，创建
skillgrade
评估。
对于结构检查使用确定性评分器，仅在需要定性判断时才使用LLM评分准则。

Step 5: Review for Hallucination Gaps

步骤5：检查幻觉漏洞

Inspect the skill for any step where the agent is forced to guess.
Replace ambiguous prose with concrete commands, file names, or output expectations.
检查技能中是否存在任何可能导致Agent被迫猜测的步骤。
将模糊的文本替换为具体的命令、文件名或输出预期。

Error Handling

错误处理

  • If a skill cannot be validated, reduce scope until the missing behavior becomes testable.
  • If the description is too broad, tighten the trigger text before adding more instructions.
  • If the supporting material grows too large, extract it into a separate file and point the agent to it explicitly.
  • If an inline command would require broad shell access or produce noisy output, keep the skill static and tell the agent to inspect the files explicitly instead.
  • 如果技能无法被验证，缩小范围直至缺失的行为可被测试。
  • 如果描述过于宽泛，在添加更多指令前先收紧触发文本。
  • 如果支持材料过大，将其提取到单独文件中，并明确引导Agent查看该文件。
  • 如果内联命令需要广泛的Shell权限或会产生大量噪声输出，保持技能静态，并告知Agent直接检查相关文件。