| name | sigil |
| description | Generates project-specific coding skills based on repository structure, tech stack, and conventions. Use when creating reusable prompts, workflows, or coding patterns tailored to a project. |
Overview
This skill generates reusable coding skills tailored to a specific project.
It analyzes:
- tech stack
- project structure
- coding conventions
Then creates focused, practical skills that improve developer productivity.
1. Analyze Before Generating
Before creating any skill:
- Identify the tech stack (frameworks, languages, tools)
- Detect project structure and patterns
- Observe naming, error handling, and testing style
Ask:
- What patterns repeat frequently?
- What tasks are complex or error-prone?
- What would benefit from reuse?
Do not generate generic skills. Always adapt to the project.
2. Generate Only High-Value Skills
Create skills only if they:
- solve a recurring problem
- reduce complexity or errors
- standardize important workflows
Avoid creating skills for:
- trivial tasks
- one-time operations
- already simple logic
Ask:
"Will this save time repeatedly?"
3. Keep Skills Small and Focused
Each skill should:
- solve one clear problem
- be easy to understand
- be immediately usable
Prefer:
- short instructions
- concrete steps
- real examples
Avoid:
- long explanations
- abstract theory
- multi-purpose skills
4. Match Project Conventions
Every generated skill must:
- follow existing naming patterns
- use the same libraries and tools
- match code style and structure
Do not introduce:
- new frameworks
- different patterns
- conflicting conventions
5. Write Clear Activation Descriptions
Descriptions must explain when to use the skill, not just what it does.
Good description:
- "Generate API routes with validation and error handling following project patterns"
Bad description:
Make descriptions:
- specific
- action-oriented
- aligned with real developer requests
6. Avoid Duplication
Before generating a skill:
- check if a similar pattern already exists
- avoid overlapping functionality
If overlap exists:
- improve the existing approach
- or merge into a single clearer skill
7. Prefer Practical Output
Every skill should produce:
- working code
- clear steps
- real usage patterns
Avoid:
- vague guidelines
- theoretical recommendations
Expected Behavior
This skill is working correctly if:
- generated skills are reused frequently
- they match the project's real patterns
- they reduce repeated effort
- they stay simple and focused
- they improve consistency across the codebase
Example Usage
User request: "Create a skill for handling API validation"
With this skill:
- Analyze existing API patterns
- Detect validation approach (e.g., Zod, Joi)
- Generate a skill using the same structure
- Ensure it matches project conventions
- Keep it simple and reusable