Part of Extend Claude With Knowledge Work Skills
Engineering Blueprint
Safety checkedBuild Skills That Extend Claude
Guide for creating modular, self-contained skills that transform Claude into a specialized agent.
What problem does this solve?
Provides structured guidance for creating reusable skills that extend Claude with specialized knowledge, workflows, and tool integrations. Skills package information into three layers (metadata, SKILL.md instructions, bundled resources) using progressive disclosure to keep context efficient while maintaining deep expertise. It covers the entire skill lifecycle: understanding use cases, planning contents, initializing structure, implementing resources, and packaging for distribution.
How does it work?
Follow a six-step process: 1. Understand the skill through concrete use-case examples. 2. Plan reusable contents (scripts, references, assets) by analyzing each example. 3. Initialize the skill directory with init_skill.py, which generates a template. 4. Edit the skill by implementing resources and writing clear SKILL.md instructions. 5. Package the skill with package_skill.py, which validates structure and creates a distributable file. 6. Iterate based on real usage feedback. Output: A packaged .skill file with metadata for triggering, concise SKILL.md instructions, and bundled resources that other instances of Claude can use immediately.
Included Skills
No skills in this group
What's required to run this?
Skills use YAML frontmatter (name, description) to trigger and can include Python, Bash, or other executables in scripts/. References files support Markdown. Assets can be any file type. Avoid deeply nested references. For large files (>100 lines), include a table of contents. No separate README.md or documentation files; only include what's necessary for Claude to execute the skill.
Key Features
Three-Layer Progressive Disclosure
Metadata (always loaded) → SKILL.md body (on trigger) → Bundled resources (as needed). Keeps context efficient.
Five-File Skill Anatomy
SKILL.md (required), scripts/ (executable code), references/ (documentation), assets/ (templates/files), and optional supporting files.
Design Principles Guidance
Concise is key (only add what Claude doesn't know), set appropriate degrees of freedom (high for flexible tasks, low for fragile operations), and progressive disclosure patterns.
Reference Organization
Patterns for multi-domain skills (split by domain), multi-framework skills (split by framework/variant), and conditional details (link to advanced content).
Automated Validation and Packaging
init_skill.py creates templates, package_skill.py validates structure and generates distributable .skill files with proper error reporting.
Tools in this Blueprint
About This Blueprint
- Industry
- Technology
More Blueprints to explore
Answer pre-purchase questions to convert hesitant shoppers with AI
Shoppers get accurate answers about fit, compatibility, ingredients, and delivery even when your team is offline, so fewer leave before checkout. Conversation insights also reveal which product-page details could prevent the next unanswered question.
Urska B.
GTM Lead
Find the root cause behind repeat support tickets with AI
Catch a few unusual tickets before a product or fulfillment issue turns into hundreds of refunds, chargebacks, and bad reviews. Cluster transcript evidence by SKU, order date, location, and carrier so operations can trace the cause and support can reach affected customers early.
Urska B.
GTM Lead