Step 01
Set Up Your Workspace
Create an empty local folder for Claude skills, initialize Git, and open the folder in Cursor. Keep generated work isolated from repositories that contain credentials or unrelated source code.
IntermediateEngineering
How to Create a Reusable 'Meta-Skill' for Generating Claude Skills with Cursor
Build a small meta-skill that creates other Claude skills consistently: ground it in the current format, keep the folder lean, validate the structure, and test whether it produces a useful second skill.
Try this with your AI assistant
Download the skill for this workflow. Version 1.0.1.
Source episode · 16:00Claude Skills Explained: My Workflow for Creating Reusable AI Agents with Cursor and Claude Code
with Claire Vo
Claire opens an empty Claude Skills folder in Cursor, supplies the official skill documentation, and creates a reusable skill-building skill with a SKILL.md file, templates, examples, and a validation script.
Before you start
What you need
- Cursor and a new local skills repository
- The current Claude Skills documentation
- One representative skill you want the meta-skill to create
- Naming, folder, and validation conventions
- A Git repository for reviewing generated files
What you’ll make
A reusable create-skill folder with a focused SKILL.md, only the supporting files it needs, and a safe validator that can check newly generated skills.
Tools used
- CursorVisitCursor
AI-first code editor
- ClaudeVisitClaude
Anthropic AI assistant
- Claude CodeVisitClaude Code
Claude for coding tasks
The workflow
Follow the sequence once, then adapt the prompts, checks, and handoffs to your own setup.
4 steps
Step 02
Initiate Skill Creation in Cursor
Ask Cursor for a meta-skill whose only job is to create maintainable Claude skills. Define the questions it must ask and the minimum files it may create.
Example prompt
Create a Claude skill named create-skill that builds other Claude skills. Before writing files, collect the target task, trigger examples, inputs, output, constraints, and required supporting resources. Default to one SKILL.md and add a reference, template, example, or validator only when it has a clear purpose.
Having a dedicated folder for all your skills keeps your workspace organized and makes it easy for tools like Claude Code to discover them.
Step 03
Prompt with Official Documentation
Attach or link the current official documentation and make it the authority for metadata, naming, file layout, references, and any validation rules. Do not rely on the model's memory of the format.
Example prompt
Use this documentation as the source of truth: [documentation]. Update create-skill to follow its required metadata and folder rules. Cite the relevant documentation section in comments or notes, and flag any instruction that the documentation does not settle.
Step 04
Review the Generated Skill
Inspect every generated file and run the read-only validator. Then use the meta-skill to create one real test skill and check its trigger behavior, output, folder contents, and diff.
Example prompt
Use create-skill to create a skill for [task]. Inputs: [inputs]. Expected output: [output]. Trigger examples: [examples]. Exclusions: [exclusions]. Generate the smallest useful folder, run the read-only validator, and list every file created.
What good looks like
- The metadata and folder layout match the supplied documentation.
- The meta-skill asks for the target task, inputs, output, examples, and constraints before generating files.
- Supporting templates and examples are referenced only when they help the task.
- A generated test skill passes the validator and works on representative input.
Build your next product with ChatPRD
Turn an idea into a PRD, user stories, and a plan.
Runbook notes
How to recover when the loop fails and where human judgment helps.
If it goes sideways
- The generated format follows remembered conventions instead of the supplied documentation
- Keep the documentation in context, cite the relevant rules in the generated SKILL.md, and recheck the structure when the platform changes.
- The meta-skill creates licenses, guides, examples, or scripts that the target skill does not need
- Make every extra file justify its runtime or maintenance value and default to one SKILL.md until a reference, template, or script is necessary.
- The generated skill description is too vague for Claude to know when to use it
- Name the concrete task, input clues, exclusions, and expected output in the description, then test both matching and non-matching requests.
- A generated validator modifies files or runs untrusted code
- Keep validation read-only, inspect scripts before execution, constrain their paths to the skill folder, and reject shell or network side effects.
Related workflows
intermediateDataMeta Muse
How to Use AI for Hobby Improvement and Data Analysis for Golf & Chess
Connect data from your hobbies (like golf scores or chess games) to an AI agent to get automated analysis, visual dashboards, and actionable coaching insights. Turn raw data into a story that helps you improve.
Open
beginnerPersonalMeta Muse
How to Create a Personalized Shopping Catalog from Your Instagram Saves Using AI
Turn your unstructured Instagram 'Saved' collection into a fun, personalized shopping catalog. This workflow uses an AI agent to organize your saves and present them in a nostalgic, engaging format like the SkyMall magazine.
Open
advancedContentMeta Muse
How to Revive an Old Blog or Digital Archive as a Personal AI Podcast
Resurrect your old digital content, like a forgotten blog from Blogger or Tumblr, by using an AI agent to find, consolidate, and transform it into a daily personal podcast. Reconnect with your past self in a new, listenable format.
Open
Start shipping better products.
Join 250,000+ product managers who use ChatPRD to write better docs, align teams faster, and build products users love.
Free to start · 14-day Pro trial · SOC 2 Type II
