letta-ai/skill-development

Skill Development

Guide for creating effective skills and contributing them to the communal knowledge base.

When to Use This Skill

Creating skills:

• Building a new skill for a specific domain or workflow
• Updating an existing skill with improvements
• Packaging skills for distribution

Contributing skills:

• You discovered something that took significant time to figure out
• Existing skill instructions were incomplete or led you astray
• You repeatedly solve the same undocumented problem

---

Part 1: Creating Skills

Skill Structure

skill-name/
├── SKILL.md (required)
│   ├── YAML frontmatter (name, description)
│   └── Markdown instructions
└── Bundled Resources (optional)
    ├── scripts/      - Executable code
    ├── references/   - Documentation loaded as needed
    └── assets/       - Files used in output

SKILL.md Requirements

---
name: skill-name
description: What this skill does. Use when [triggers]. 
---

# Skill Name

[Instructions in imperative form...]

Metadata rules:

• name: lowercase, hyphens, gerund form (e.g., creating-skills)
• description: Lead with action verbs, end with "Use when [triggers]"

Writing Style

Write in imperative/infinitive form (verb-first), not second person:

• ✅ "To accomplish X, do Y"
• ❌ "You should do X"

Progressive Disclosure

Keep SKILL.md lean (<5k words). Move detailed content to references/:

1. Metadata (~100 words) - Always in context
2. SKILL.md body (<5k words) - When skill triggers
3. References (unlimited) - Loaded as needed by agent

Scripts

Include scripts for tasks that:

• Are repeatedly rewritten
• Require deterministic reliability
• Save significant tokens

---

Part 2: Contributing Skills

Core Philosophy

This repository is a living knowledge base. Skills must contain general-purpose knowledge that helps many agents - not project-specific configs or personal preferences.

What to Contribute

| ✅ Contribute | ❌ Don't Contribute | |--------------|-------------------| | Patterns that appear 3+ times | One-off solutions | | General API/tool patterns | Project-specific configs | | Battle-tested workarounds | Personal preferences | | Gap in existing skills | Already documented elsewhere |

Validation Checklist

Before contributing, verify:

• [ ] Is this generalizable beyond your specific context?
• [ ] Have you seen this pattern multiple times?
• [ ] Does this address a real gap vs. personal preference?
• [ ] Would this help an agent on a completely different project?

Contribution Process

1. Recognize - Notice patterns during work (time investment, repetition, corrections)
2. Validate - Confirm pattern is general, not specific to your context
3. Create branch - add/skill-name or fix/skill-name
4. Submit PR - Clear description with rationale and evidence

See references/pr-workflow.md for detailed PR process.

---

Scripts

• scripts/init_skill.py - Initialize skill directory structure
• scripts/package_skill.py - Package skill for distribution
• scripts/quick_validate.py - Validate SKILL.md format

References

• references/recognizing-learnings.md - Patterns for spotting valuable learnings
• references/validation-criteria.md - Detailed validation guidelines
• references/pr-workflow.md - PR process and templates
• references/contribution-examples.md - Real contribution examples
• references/progressive-disclosure-research.md - Research on skill organization