| name | implement-agent-skills |
| description | Comprehensive workflow for creating, implementing, and validating Agent Skills. Use when asked to "create a new skill", "author a skill", "add a capability", or when standardizing project-specific workflows. Support for platform detection (Cursor, Claude Code, Gemini CLI, Codex) and template selection. |
Implement Agent Skills
This skill provides a structured process for creating effective Agent Skills that extend agent capabilities with specialized knowledge, workflows, and tools.
Platform Setup
Before starting, identify the target AI coding tool environment to ensure compatibility.
Skill Creation Process
Follow these steps to develop a new skill from scratch or improve an existing one.
1. Understand the Skill with Concrete Examples
- Clarify the functionality the skill should support.
- Identify what a user would say to trigger this skill.
- Examples: "How many users logged in today?", "Rotate this PDF", "Build me a dashboard".
2. Brainstorm and Evaluate Approaches
- Broadly and deeply think of five different approaches to implement the skill.
- Score each approach (0 to 100) based on Feasibility, Performance, Maintainability, and Complexity.
- Recommend the best approach and show clear reasons why it was chosen.
- Proceed with the recommended approach unless the user explicitly wants to select a different one.
- Specifically, the PLAN mode MUST include the decision process.
3. Plan Reusable Skill Contents
Analyze the examples to identify reusable resources:
- Scripts: For tasks requiring deterministic reliability.
- References: For documentation, schemas, or domain knowledge.
- Assets: For templates, boilerplate code, or static files.
4. Initialize the Skill
- Create the skill directory following the naming standards.
- Create the
SKILL.md file with required YAML frontmatter.
- Use an appropriate template from the template catalog.
5. Edit the Skill
- Implement Resources: Build out
scripts/, references/, and assets/.
- Write Instructions: Fill in the
SKILL.md body using imperative language.
- Apply Progressive Disclosure: Keep
SKILL.md lean by linking to references.
6. Package the Skill
- Ensure all file references are relative and only one level deep.
- Validate the YAML frontmatter and directory structure.
- (Optional) Use a packaging script if available in the project.
7. Iterate
- Test the skill on real-world tasks.
- Refine descriptions to improve discovery.
- Update instructions based on edge cases discovered during testing.
Resources