AgentSkillsCN

gemini-skill-creator

为 Antigravity 代理环境创建结构化的技能。适用于当用户请求创建新技能、构建技能体系，或提及代理能力的技能开发时使用。

SKILL.md
--- frontmatter
name: gemini-skill-creator
description: Creates structured skills for the Antigravity agent environment. Use when the user requests to create a new skill, build a skill system, or mentions skill development for agent capabilities.

Gemini Skill Creator

When to use this skill

  • •User asks to create a new skill for the Antigravity agent
  • •User mentions building agent capabilities or extending agent functionality
  • •User requests skill templates or skill structure guidance
  • •User wants to standardize skill creation across projects

Core Structural Requirements

Every skill you generate must follow this folder hierarchy:

code
<skill-name>/
├── SKILL.md          (Required: Main logic and instructions)
├── scripts/          (Optional: Helper scripts)
├── examples/         (Optional: Reference implementations)
└── resources/        (Optional: Templates or assets)

YAML Frontmatter Standards

The SKILL.md must start with YAML frontmatter following these strict rules:

  • •name: Gerund form (e.g., testing-code, managing-databases). Max 64 chars. Lowercase, numbers, and hyphens only. No "claude" or "anthropic" in the name.
  • •description: Written in third person. Must include specific triggers/keywords. Max 1024 chars.
    • •Example: "Extracts text from PDFs. Use when the user mentions document processing or PDF files."

Writing Principles (The "Claude Way")

When writing the body of SKILL.md, adhere to these best practices:

Conciseness

  • •Assume the agent is smart
  • •Do not explain basic concepts (what a PDF is, what a Git repo is)
  • •Focus only on the unique logic of the skill

Progressive Disclosure

  • •Keep SKILL.md under 500 lines
  • •If more detail is needed, link to secondary files (e.g., [See ADVANCED.md](ADVANCED.md))
  • •Only one level deep for references

Path Standards

  • •Always use / for paths, never \

Degrees of Freedom

Choose the right format based on task flexibility:

  • •Bullet Points: High-freedom tasks (heuristics, decision-making)
  • •Code Blocks: Medium-freedom (templates, structured examples)
  • •Specific Bash Commands: Low-freedom (fragile operations, exact sequences)

Workflow & Feedback Loops

For complex tasks, include:

  1. •Checklists: A markdown checklist the agent can copy and update to track state
  2. •Validation Loops: A "Plan-Validate-Execute" pattern
    • •Example: Run a script to check a config file BEFORE applying changes
  3. •Error Handling: Instructions for scripts should be "black boxes"
    • •Tell the agent to run --help if they are unsure

Skill Creation Workflow

When creating a new skill, follow these steps:

1. Understand Requirements

  • • Clarify the skill's purpose
  • • Identify trigger keywords/scenarios
  • • Determine required tools/scripts
  • • Define expected inputs and outputs

2. Design Structure

  • • Choose appropriate folder structure
  • • Determine if scripts/examples/resources are needed
  • • Plan the workflow steps

3. Write SKILL.md

  • • Create YAML frontmatter with proper name and description
  • • Add "When to use this skill" section
  • • Document the workflow with checklist
  • • Write clear, concise instructions
  • • Link to supporting files if needed

4. Create Supporting Files

  • • Add scripts to scripts/ if needed
  • • Provide examples in examples/ if helpful
  • • Include templates/assets in resources/ if applicable

5. Validate

  • • Check YAML frontmatter follows standards
  • • Verify skill name is in gerund form
  • • Ensure description is third-person with triggers
  • • Confirm SKILL.md is under 500 lines
  • • Test that paths use forward slashes

Output Template

When creating a skill, use this structure:

markdown
---
name: [gerund-name]
description: [3rd-person description with triggers]
---

# [Skill Title]

## When to use this skill
- [Trigger 1]
- [Trigger 2]

## Workflow
- [ ] [Step 1]
- [ ] [Step 2]
- [ ] [Step 3]

## Instructions
[Specific logic, code snippets, or rules]

## Resources
- [Link to scripts/ or resources/]

Example Skill Names

Good examples:

  • •testing-code
  • •managing-databases
  • •deploying-applications
  • •processing-pdfs
  • •analyzing-logs

Bad examples:

  • •test (not gerund)
  • •DatabaseManager (not lowercase)
  • •claude-helper (contains "claude")
  • •do_deployment (uses underscore)

Location Guidelines

Project-Specific Skills

Store in: .agent/skills/<skill-name>/

Global Skills

Store in: ~/.agent/skills/<skill-name>/

Global skills are available across all projects and should be used for:

  • •Universal development workflows
  • •Cross-project utilities
  • •Standardized processes
  • •Reusable templates

Best Practices

  1. •Start Simple: Begin with minimal structure, add complexity only when needed
  2. •Be Specific: Include exact commands for critical operations
  3. •Provide Context: Add "When to use" sections with clear triggers
  4. •Enable Self-Service: Include --help references for scripts
  5. •Maintain Consistency: Follow naming and structure conventions
  6. •Think Reusability: Design skills to work across different projects
  7. •Document Assumptions: State prerequisites clearly
  8. •Test Instructions: Ensure steps are reproducible

Common Patterns

For Testing Skills

  • •Include setup/teardown steps
  • •Provide example test cases
  • •Document assertion patterns

For Deployment Skills

  • •Add validation checkpoints
  • •Include rollback procedures
  • •Document environment requirements

For Processing Skills

  • •Define input/output formats
  • •Include error handling
  • •Provide sample data

Error Handling

When scripts fail:

  1. •Check script exists and is executable
  2. •Run with --help to understand usage
  3. •Verify prerequisites are met
  4. •Check environment variables
  5. •Review error messages for specific issues

References

  • •Original skill creation guide: /home/bamboosam/teddy/skillcreateinfo.md
  • •Skills directory: .agent/skills/ (project) or ~/.agent/skills/ (global)