skill-forge-build

Skill Builder — Scaffold & Generate

Process

Step 1: Gather Inputs

Determine what we're building from:

• Plan document: Output from /skill-forge plan (preferred)
• Description: Natural language description of the skill
• Existing skill: Path to a skill to use as foundation

If no plan exists, run a quick planning pass (ask 3 key questions):

1. What does the skill do? (1-2 sentences)
2. What commands should it have? (list)
3. Does it need scripts or external tools? (yes/no)

Step 2: Generate Frontmatter

The most critical step. The description field determines activation.

Framework for writing descriptions:

[Capability statement] + [Detailed capabilities] + [Trigger phrases]

Rules:

• Under 1024 characters total
• No XML angle brackets (< >)
• Include 5-10 trigger phrases users would say
• Mention file types if relevant
• Add negative triggers if risk of over-triggering

Use references/description-guide.md for the full framework.

Step 3: Write Main SKILL.md

Structure the body following this template:

# [Skill Name] -- [Tagline]

[1-2 sentence overview]

## Quick Reference

| Command | What it does |
|---------|-------------|
| /name cmd1 | Description |
| /name cmd2 | Description |

## Orchestration Logic

[How commands route to sub-skills]
[Decision trees for conditional routing]

## [Domain-Specific Section]

[Core rules, thresholds, quality gates]

## Reference Files

[List of on-demand reference files]

## Sub-Skills

[List of sub-skills with one-line descriptions]

Critical SKILL.md rules:

• Under 500 lines
• Under 5000 tokens
• Actionable instructions (not vague)
• Include error handling
• Link references, don't inline them

Step 4: Generate Sub-Skills (Tier 3-4)

For each sub-skill, generate a SKILL.md with:

---
name: parent-child
description: >
  [What it does]. Use when user says "[trigger 1]", "[trigger 2]",
  "[trigger 3]", or "[trigger 4]".
---

Sub-skill body guidelines:

• Self-contained instructions for ONE workflow
• Cross-reference other sub-skills by name (not inline their content)
• Reference shared files in the parent skill's references/ directory
• Include explicit input/output definitions