Publishing Skills

Share your expertise with the Hubify community by publishing skills. Well-crafted skills help thousands of agents worldwide.

Skill Structure

Minimum Viable Skill

# my-skill/.hub
name: my-skill
version: 0.1.0
category: coding
description: A brief description of what this skill does
prompt: |
  Your skill instructions here...

Complete Skill

name: typescript-strict-mode
version: 1.0.0
category: coding

author: "@yourusername"
license: MIT

description: |
  Enforce strict TypeScript configuration for improved type safety

platforms:
  - claude-code
  - cursor
  - windsurf

use_cases:
  - Configure tsconfig.json for strict mode
  - Add strict type checking rules
  - Enforce null checks

tags:
  - typescript
  - configuration
  - type-safety

dependencies:
  requires: []
  extends: []
  integrates_with:
    - eslint-typescript

prompt: |
  When setting up TypeScript strict mode, follow these steps:

  1. **Enable strict mode** in tsconfig.json:
     ```json
     {
       "compilerOptions": {
         "strict": true,
         "noUncheckedIndexedAccess": true,
         "exactOptionalPropertyTypes": true
       }
     }

2. Handle existing errors by:

   • Starting with strict: false if you have many errors
   • Enabling strict flags incrementally
   • Using // @ts-expect-error sparingly for legacy code

3. Configure ESLint integration for runtime checks

examples:

• title: "New Next.js Project" prompt: "Set up strict TypeScript for my new Next.js app" expected: "Configure tsconfig.json with strict mode enabled"

• title: "Existing Project Migration" prompt: "Migrate my existing project to strict TypeScript" expected: "Incremental migration strategy with error handling"

## Creating Skills

### Method 1: Generate from Prompt

```bash
hubify generate "Python code review checklist with security focus"

This uses AI to create a skill scaffold:

Generating skill from prompt...

Created: python-security-review/.hub

  Name: python-security-review
  Category: security
  Use Cases:
    - Review Python code for security issues
    - Check for injection vulnerabilities
    - Validate input handling

Edit the .hub file to refine the skill, then publish.

Method 2: Create Manually

Create a directory with a .hub file:

mkdir my-skill
cd my-skill

Create .hub:

name: my-skill
version: 0.1.0
category: coding
description: What this skill does
prompt: |
  Your instructions...

Method 3: Fork Existing Skill

hubify fork typescript-strict-mode --name my-typescript-config

This creates a local copy you can modify and publish.

Skill Categories

Category	Description
coding	Programming patterns and practices
workflow	Development workflow automation
documentation	Documentation generation
testing	Testing strategies
security	Security best practices
devops	Deployment and infrastructure
research	Research and analysis

Writing Effective Prompts

Be Specific

# Bad
prompt: "Help with TypeScript"

# Good
prompt: |
  When configuring TypeScript strict mode:
  1. Enable `strict: true` in tsconfig.json
  2. Add `noUncheckedIndexedAccess` for safer array access
  3. Enable `exactOptionalPropertyTypes` for precise optional handling

Include Examples

prompt: |
  ...

  Example tsconfig.json:
  ```json
  {
    "compilerOptions": {
      "strict": true,
      "noUncheckedIndexedAccess": true
    }
  }

### Handle Edge Cases

```yaml
prompt: |
  ...

  For existing projects with many type errors:
  1. Start with `strict: false`
  2. Enable flags incrementally:
     - Week 1: `strictNullChecks`
     - Week 2: `strictFunctionTypes`
     - Week 3: Full `strict: true`

Validating Skills

Validate Locally

hubify hub validate ./my-skill

Output:

Validating my-skill...

  ✓ Valid .hub file
  ✓ Name follows conventions
  ✓ Category is valid
  ✓ Prompt is non-empty
  ✓ Version is semver

Ready to publish!

Test in Sandbox

hubify test my-skill

Runs the skill in an E2B sandbox to verify it works.

Publishing

Publish to Registry

hubify publish ./my-skill

Output:

Publishing my-skill...

  Validating... ✓
  Uploading... ✓
  Indexing... ✓

Published: my-skill@0.1.0

View at: https://hubify.com/skills/my-skill

Publish Options

# Dry run (preview without publishing)
hubify publish ./my-skill --dry-run

# Publish as unlisted (not in search)
hubify publish ./my-skill --unlisted

# Publish with specific tag
hubify publish ./my-skill --tag beta

Versioning

Follow semantic versioning:

Change Type	Version Bump	Example
Bug fixes, typos	Patch	1.0.0 → 1.0.1
New examples, clarifications	Minor	1.0.0 → 1.1.0
Core prompt changes	Major	1.0.0 → 2.0.0

Publishing Updates

# Edit .hub, bump version
# version: 0.1.0 → version: 0.2.0

hubify publish ./my-skill

Trust & Verification

New skills start with low confidence and build trust through:

1. Executions: More agents using the skill
2. Success Rate: High success percentage
3. Agent Diversity: Used across different platforms
4. Evolution Stability: Improvements over time

Verification Levels

Level	Requirements
Unverified	Just published
Community	100+ executions, 80%+ success
Verified	1000+ executions, 90%+ success, manual review
Official	Hubify team maintained

Skill Maintenance

Monitor Performance

hubify learn stats my-skill

View Improvement Suggestions

hubify evolve status my-skill

Accept Evolutions

# View pending evolution
hubify evolve preview my-skill

# Accept evolution
hubify evolve accept my-skill

Best Practices

Do

• Write clear, actionable instructions
• Include real examples
• Handle common edge cases
• Test on multiple platforms
• Respond to improvement suggestions

Don't

• Publish duplicate skills
• Include sensitive information
• Make prompts too generic
• Ignore user feedback

Monetization

Coming soon: Premium skills marketplace with revenue sharing for authors.

Next Steps