opencode-skill-generator

Overview

Transform raw prompts, rules, or documentation into high-density SKILL.md files following Opencode standards.

Workflow

1. Analysis & Extraction

Analyze input to extract:

• Intent: What task does this skill solve?
• Triggers: Specific words or file globs.
• Audience: Who is this for? (e.g., developers, writers).
• Logic: The step-by-step workflow.

2. Frontmatter Generation

Construct the YAML header:

• name: kebab-case (e.g., git-commit-gen).
• description: < 120 chars, action-oriented.
• compatibility: opencode.
• metadata: Map audience, language, scope, and triggers.

3. Content Optimization

Rewrite the logic using Token-Efficient Principles:

• Imperative Mood: “Check file” (not “You should check the file”).
• Tabular Data: Use tables for mappings or key-value pairs.
• Condensation: Remove conversational filler.
• Structure:
  1. ## Overview: 1-sentence summary.
  2. ## Configuration (Optional): Setup/Defaults.
  3. ## Workflow or ## Rules: Core logic.

4. Output Template

Generate the file using this exact structure:

---

name: [skill-name]
description: [Short description]
license: MIT
compatibility: opencode
metadata:
audience: [target]
triggers: [list]

---

## Overview

[Concise summary]

## Workflow

### 1. Trigger

[When to run]

### 2. [Step Name]

[Actionable instruction]

- [Detail 1]
- [Detail 2]

| Input | Output |
| :---- | :----- |
| [A]   | [B]    |