- Home
- Skills
- Notedit
- Happy Skills
- Skill Creation Guide
skill-creation-guide_skill
- Python
278
GitHub Stars
2
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill notedit/happy-skills --skill skill-creation-guide- LICENSE.txt11.1 KB
- SKILL.md17.4 KB
Overview
This skill is a practical guide for creating effective skills that extend Claude with domain knowledge, workflows, and tool integrations. It explains what to include, how to structure reusable resources, and how to keep the agent context efficient and maintainable. Use it to design new skills or improve existing ones with clear, repeatable patterns.
How this skill works
The guide explains the anatomy of a skill, recommended resource types (scripts, references, assets), and a progressive disclosure loading model that keeps context small. It walks through a step-by-step creation process: understand concrete examples, plan reusable contents, initialize the skill scaffold, implement resources, and iterate based on real usage. It also gives guidance on degrees of freedom and when to prefer concise instructions versus precise scripts.
When to use it
- Creating a new skill that automates a repeatable workflow or integrates a tool/API
- Refactoring an existing skill to reduce context bloat and improve discoverability
- Designing reusable scripts, reference docs, or output assets for frequent tasks
- Deciding how much specificity a skill's instructions should include (high/medium/low freedom)
- Organizing content so Claude only loads detailed references when needed
Best practices
- Keep the in-context guidance minimal—only include information Claude truly needs
- Organize heavy or long content into separate reference files and link them clearly
- Classify instructions by freedom level: text guidance, pseudocode, or concrete scripts
- Use assets for files meant to be copied into outputs and scripts for deterministic tasks
- Prefer one-level references from the main guide and include a TOC for long files
Example use cases
- A PDF processing skill with a scripts/rotate_pdf.py for deterministic rotations
- A BigQuery skill that stores table schemas in references/schema.md for quick lookup
- An image editor skill bundling templates in assets/ and precise scripts for fragile ops
- A cloud deploy skill that keeps provider-specific docs in separate reference files
- A frontend boilerplate skill providing a reusable assets/hello-world template
FAQ
Put core workflow steps and trigger criteria in the main guide; move large schemas, full API docs, and lengthy examples into reference files linked from the guide.
When should I include executable scripts?
Include scripts for operations that repeat often or must be deterministic and error-free; use scripts to avoid re-sending the same code in context.