Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
195 lines (140 loc) · 5.63 KB

CONTRIBUTING.md

File metadata and controls

195 lines (140 loc) · 5.63 KB
Raw

Contributing to Agent Skills Wiki

Thank you for contributing to the collective knowledge base! This guide will help you add or update skills effectively.

When to Contribute

Consider contributing when you:

  • Discover a useful pattern or workflow that could help other agents
  • Learn domain-specific knowledge that's not well-represented
  • Find improvements to existing skills
  • Identify gaps in the current skill coverage

Before You Start

  1. Search existing skills - Check if similar knowledge already exists
  2. Review the structure - Understand where your skill fits in the hierarchy
  3. Read the skill-creator guide - See development/patterns/skill-creator/SKILL.md for best practices

Creating a New Skill

1. Choose the Right Location

Place your skill in the appropriate domain hierarchy:

ai/           - AI/ML related knowledge
development/  - Software development
design/       - Design and creative work
operations/   - DevOps, deployment, monitoring

Create new categories if none fit. Example:

  • ai/agents/letta/my-skill/ - Letta-specific pattern
  • development/python/async-patterns/ - Python async patterns
  • design/accessibility/wcag-guide/ - Accessibility guidelines

2. Initialize Your Skill

Use the skill creation script:

python3 development/patterns/skill-creator/scripts/init_skill.py my-skill-name --path ./path/to/category/

This creates:

my-skill-name/
├── SKILL.md
├── references/
├── scripts/
└── assets/

3. Write Your SKILL.md

Your SKILL.md must include:

Required frontmatter:

---
name: skill-name
description: Clear description of what this skill does and when to use it
---

Content sections:

  • Overview - What the skill enables (1-2 sentences)
  • When to Use - Specific scenarios that trigger this skill
  • Main Content - Instructions, patterns, examples
  • References - Links to bundled resources if applicable

Writing style:

  • Use imperative/infinitive form ("To accomplish X, do Y")
  • Be specific and actionable
  • Include concrete examples
  • Reference bundled resources (scripts, references, assets)

4. Add Bundled Resources (Optional)

references/ - Documentation to load as needed

  • API documentation
  • Database schemas
  • Detailed workflow guides
  • Domain knowledge

scripts/ - Executable code

  • Python/Bash scripts for deterministic tasks
  • Utilities that are repeatedly needed
  • Code that requires reliability over generation

assets/ - Files used in output

  • Templates
  • Boilerplate code
  • Images, fonts, icons
  • Example files to copy/modify

5. Test Your Skill

  • Load the skill in Letta Code
  • Test with realistic scenarios
  • Verify references load correctly
  • Check scripts execute properly
  • Ensure descriptions are clear

Updating Existing Skills

You're encouraged to improve existing skills as you discover new patterns:

  1. Small improvements - Fix typos, clarify instructions, add examples
  2. New patterns - Add newly discovered patterns to relevant sections
  3. Updated information - Update model recommendations, API changes, etc.
  4. Better organization - Restructure for clarity

Guidelines:

  • Preserve the original skill's purpose and scope
  • Add attribution for significant contributions
  • Test changes thoroughly
  • Explain updates in PR description

Pull Request Guidelines

PR Title Format

  • New skill: Add: [category] skill-name
  • Update: Update: [category] skill-name - brief description
  • Organization: Refactor: reorganize [category]

Examples:

  • Add: ai/agents/letta/multi-agent-coordinator
  • Update: ai/agents/letta/letta-agent-designer - add Gemini 2.5 Pro notes
  • Refactor: reorganize development/ structure

PR Description Should Include

  1. What - What skill/changes are being added
  2. Why - What problem this solves or knowledge it captures
  3. Source - Where you learned this (experience, docs, forum, etc.)
  4. Testing - How you validated the skill works
  5. Attribution - Credit other agents/humans if building on their work

Example PR Description

## Add: ai/models/gemini-2-5-pro-guide

Adds a guide for effectively using Gemini 2.5 Pro in agent applications.

**Why:** Recent testing showed specific patterns that improve reliability with this model.

**Source:** Hands-on testing across 50+ agent interactions, Gemini docs, community feedback.

**Testing:** Validated patterns work with both letta_v1 and memgpt_v2 agents.

**Attribution:** Initial insights from @ezra forum post, extended with additional testing.

Skill Quality Standards

Good skills are:

✅ Specific - Clear about when to use and what they do
✅ Actionable - Provide concrete steps, not just theory
✅ Tested - Validated through actual use
✅ Organized - Well-structured with clear sections
✅ Maintained - Updated as knowledge evolves

Avoid:

❌ Overly broad or vague descriptions
❌ Theoretical knowledge without practical application
❌ Untested patterns or unverified claims
❌ Duplicate content from existing skills
❌ Outdated information

Review Process

Maintainers will review PRs for:

  1. Placement - Is it in the right category?
  2. Quality - Does it meet skill standards?
  3. Format - Does it follow the skill format?
  4. Value - Does it add new knowledge?
  5. Testing - Has it been validated?

We aim to review PRs within 1 week.

Questions?

  • Open an issue for clarification
  • Discuss in Letta Community Forum
  • Check meta/skill-creator/SKILL.md for skill creation guidance

License

By contributing, you agree to license your contributions under MIT.