/skill-maker

name: skill-maker
description: >
  [production-grade internal] Creates reusable Claude Code skills and plugins
  when you want to automate repeatable workflows into shareable tools.
  Routed via the production-grade orchestrator.

Skill Maker

Protocols

!`cat Claude-Production-Grade-Suite/.protocols/ux-protocol.md 2>/dev/null || true` !`cat Claude-Production-Grade-Suite/.protocols/input-validation.md 2>/dev/null || true` !`cat Claude-Production-Grade-Suite/.protocols/tool-efficiency.md 2>/dev/null || true` !`cat Claude-Production-Grade-Suite/.protocols/visual-identity.md 2>/dev/null || true` !`cat Claude-Production-Grade-Suite/.protocols/freshness-protocol.md 2>/dev/null || true` !`cat Claude-Production-Grade-Suite/.protocols/receipt-protocol.md 2>/dev/null || true` !`cat Claude-Production-Grade-Suite/.protocols/boundary-safety.md 2>/dev/null || true` !`cat Claude-Production-Grade-Suite/.protocols/loop-protocol.md 2>/dev/null || true` !`cat Claude-Production-Grade-Suite/.protocols/conflict-resolution.md 2>/dev/null || true` !`cat Claude-Production-Grade-Suite/.protocols/platform-adaptation.md 2>/dev/null || true`

**Portability:** the `` !` `` lines above are preamble commands; if your host did not auto-execute them, run them yourself. If any tool this skill names is missing in your environment, apply `platform-adaptation.md` (workspace `.protocols/` copy, or `skills/_shared/` in this plugin's repo) — same guarantees, host-native mechanisms.

!`cat .production-grade.yaml 2>/dev/null || echo "No config — using defaults"`

**Fallback (if protocols not loaded):** Use AskUserQuestion with options (never open-ended), "Chat about this" last, recommended first. Work continuously. Print progress constantly. Validate inputs before starting — classify missing as Critical (stop), Degraded (warn, continue partial), or Optional (skip silently). Use parallel tool calls for independent reads. Use Glob before Read to map file structure. No oracle, no loop — never iterate without an executable exit check; never edit or weaken tests you must pass (tests/ is QA-owned).

Progress Output

Follow `Claude-Production-Grade-Suite/.protocols/visual-identity.md`. Print structured progress throughout execution.

**Skill header** (print on start):

━━━ Skill Maker ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

**Phase progress** (print during execution):

  [1/3] Pattern Analysis
    ✓ {N} recurring patterns identified
    ⧖ analyzing workflow structure...
    ○ skill generation
    ○ installation

  [2/3] Skill Generation
    ✓ {N} custom skills drafted
    ⧖ writing SKILL.md files...
    ○ installation

  [3/3] Installation
    ✓ {N} skills installed to .claude/skills/

**Completion summary** (print on finish — MUST include concrete numbers):

✓ Skill Maker    {N} project-specific skills created    ⏱ Xm Ys

Overview

End-to-end skill and plugin creation pipeline. Interviews the user on what the skill should do, writes the SKILL.md, packages it as a Claude Code plugin, creates a GitHub repo, and adds it to the user's marketplace — all in one flow.

Config Paths

Read `.production-grade.yaml` at startup if available. Skill-maker is mostly self-contained and does not depend on project-level path overrides.

When to Use

• User asks to create a new skill or plugin
• User describes a reusable workflow that should be a skill
• User says "make a skill", "build a plugin", "I need a skill for..."
• NOT for: editing existing skills (just edit the file directly)

Process Flow

digraph skill_maker {
    rankdir=TB;

    "Skill idea received" [shape=doublecircle];
    "Phase 1: Interview" [shape=box];
    "Phase 2: Write SKILL.md" [shape=box];
    "User approves?" [shape=diamond];
    "Phase 3: Package as plugin" [shape=box];
    "Phase 4: Create repo & push" [shape=box];
    "Phase 5: Add to marketplace" [shape=box];
    "Done" [shape=doublecircle];

    "Skill idea received" -> "Phase 1: Interview";
    "Phase 1: Interview" -> "Phase 2: Write SKILL.md";
    "Phase 2: Write SKILL.md" -> "User approves?";
    "User approves?" -> "Phase 2: Write SKILL.md" [label="revise"];
    "User approves?" -> "Phase 3: Package as plugin" [label="approved"];
    "Phase 3: Package as plugin" -> "Phase 4: Create repo & push";
    "Phase 4: Create repo & push" -> "Phase 5: Add to marketplace";
    "Phase 5: Add to marketplace" -> "Done";
}

Phase 1: Interview (Quick)

**Engagement mode determines interview depth:**

• **Express**: Infer all answers from the project context and agent prompts. Skip interview entirely. Report inferences.
• **Standard**: Ask 1-2 key questions (purpose + trigger conditions). Auto-resolve the rest.
• **Thorough/Meticulous**: Ask 3-4 questions using AskUserQuestion, one at a time:

1. **What does this skill do?** — Core purpose in one sentence 2. **When should it trigger?** — Specific words, patterns, or situations 3. **What's the workflow?** — Steps the skill should follow (linear, loop, decision tree?) 4. **Skill type?** — Options: Technique (steps to follow), Pattern (mental model), Reference (docs/API guide), Workflow (multi-phase process)

Phase 2: Write SKILL.md

Follow these rules from the writing-skills methodology:

**Frontmatter:**

• `name`: kebab-case, letters/numbers/hyphens only
• `description`: Start with "Use when...", max 500 chars, triggering conditions only — NEVER summarize the workflow

**Structure:**

---
name: skill-name
description: Use when [triggering conditions]
---

# Skill Name

## Overview
Core principle in 1-2 sentences.

## When to Use
Bullet list with symptoms and use cases.

## Process Flow (if multi-step)
Small inline dot flowchart for non-obvious decisions.

## [Core Content]
Steps, patterns, or reference material.

## Common Mistakes
Table of mistake -> fix pairs.

**Quality rules:**

• One excellent example beats many mediocre ones
• Flowcharts ONLY for non-obvious decision points
• Keep under 500 words