Skip to content

Home

Jump to bottom
Alireza Rezvani edited this page Oct 29, 2025 · 3 revisions

Claude Code Skills & Agents Factory

Welcome to the Claude Code Skill Factory "The first Claude Code and Code AI Agent Skill Creator Toolkit" wiki!

One toolkit to generate production-ready Claude Skills, Claude Code Agents, Slash Commands, and LLM Prompts at scale. Templates, examples, and guided flows for fast, reliable AI capability building across Claude apps, Claude Code, and the API.

Repo: alirezarezvani/claude-code-skills-factory Status: Active • v1.3.0 • Last updated: Oct 29, 2025

Why this Factory?

  • Ship faster: Generate complete, production-grade Skills and Agents with zero boilerplate.
  • Scale cleanly: Composable architecture patterns; consistent YAML + Python conventions.
  • Stay safe: Strict validation (naming, permissions, structure) and clear input/output contracts.
  • Works everywhere: Claude Desktop, Claude Code, Apps (browser), and API.

Great idea—here’s a concise Vision & Mission section you can drop into the Wiki page:

Vision & Mission

Vision Build an adaptive toolkit that identifies a company’s real needs and automatically composes all required -Claude Code resources Agents & Sub-agents, Agent Skills (also available for Claude AI Desktop App), Slash Commands, Hooks, MCP integrations, and built-in features —fully customized to any purpose, team and organization. -No predefined workflows: the user states the need, and the Claude Skill Factory generates the capability.

Mission Deliver an almost-automated agentic workforce that empowers individuals and companies to build their dream products—a true AI assistant at every level and scale.

What you can build

  • Claude Skills (multi-file capabilities: SKILL.md + Python + samples)
  • Claude Code Agents (single-file specialists with enhanced YAML frontmatter)
  • Slash Commands for Claude Code (official Anthropic patterns)
  • Production-ready Prompts (multi-format; XML/Claude/ChatGPT/Gemini)

New: Interactive navigation with 4 guide agents + 5 slash commands for a fully guided build flow.

Quick Links

Feature Highlights

  • Production-ready output: Proper YAML, kebab-case names, typed Python, safe ops, focused samples.
  • Composable design: Chain skills (extract → analyze → report → format) without duplication.
  • Smart generation: Determines prompt-only vs. Python, multi-file needs, and sample formats.
  • Validation & safety: Four-layer checks for command names, bash permissions, arguments, and YAML.
  • Guided build UX: 4 guide agents + 5 slash commands streamline end-to-end creation and testing.

Architecture at a glance

  • Skills → Multi-file, structured capability packs: SKILL.md (YAML + docs) • Python modules • sample_input.json • expected_output.json • HOW_TO_USE.md
  • Agents → Single .md with enhanced YAML (name/description/tools/model/color/field/expertise/MCP), plus a precise system prompt.
  • Slash Commands → Official patterns: Simple, Multi-Phase, Agent-Style.
  • Prompts → Prompt Factory with 69 presets, multi-format output, 7-point quality gates.

Getting Started (5 minutes)

1) Install a ready-made Skill/Factory

# Example: Prompt Factory
cp -r generated-skills/prompt-factory ~/.claude/skills/
# Or import the ZIP in Claude Desktop (see repo folder for .zip)

2) Generate a new Claude Skill

  • Open SKILLS_FACTORY_PROMPT.md, fill the variables, and paste into Claude (Desktop/Code/API).
  • You’ll receive a full skill package: SKILL.md, Python files (if needed), samples, HOW_TO_USE.md, and a ready-to-import ZIP.

3) Generate a Claude Code Agent

  • Open AGENTS_FACTORY_PROMPT.md, set AGENT_NAME, TOOLS, MODEL, etc.
  • Paste into Claude → receive a single .md file → place it in:
# Project-level
cp my-agent.md .claude/agents/
# User-level
cp my-agent.md ~/.claude/agents/

4) Use the Interactive Navigation System 

"I want to build something"
  • Orchestrator (factory-guide) routes to specialists (skills-guide, prompts-guide, agents-guide).
  • Answer 3–6 questions → validated output → install → test.

Included Examples (ready to run)

  • AWS Solution Architect → IaC, cost optimization, serverless patterns
  • Content Trend Researcher → Multi-platform trend + intent + outline generation
  • Microsoft 365 Tenant Manager → Config, security, PowerShell generation
  • Psychology Advisor → CBT tooling and wellness protocols
  • Agent Factory → Generate new Claude Code agents programmatically
  • Prompt Factory → 69 presets, XML/Claude/ChatGPT/Gemini formats

Browse: generated-skills/ • claude-skills-examples/

Slash Command Factory (v2.0)

Best Practices (quick checklist)

  • One purpose per Skill/Agent; keep responsibilities tight.
  • Composable IO; design outputs to feed downstream skills.
  • YAML first; kebab-case names, concise descriptions.
  • Type-safe Python; explicit error handling and clear docstrings.
  • Minimal samples; just enough to validate behavior.
  • Use the validators; let the factory catch structure/permission issues early.

Compatibility

  • Claude: Desktop, Code, Apps (browser), API
  • Models: Tuned for Claude Sonnet; configurable per Agent/Skill
  • MCP: Supported via mcp_tools in Agent YAML

FAQ

Q: Skills vs Agents—when to choose which? Use Skills for multi-file capabilities (logic + docs + samples). Use Agents when a single specialist with clear tool access and behavior is enough.

Q: Do I need Python for every Skill? No. The factory decides when prompt-only is sufficient and when typed Python modules add value.

Q: Can I extend generated content safely? Yes—follow the conventions (kebab-case, typed Python, docstrings). Add modules like validate_*.py / format_*.py as complexity grows.

Q: How do I keep Slash Commands safe? Only allow specific bash subcommands; never wildcard. Use $ARGUMENTS placeholders, and run the provided validator.

Contributing

  • Fork → add example skills to claude-skills-examples/ → ensure formatting + samples → PR.
  • See contribution notes in the main README.md.

License

This repo offers example templates and generation systems. Skills and agents you generate are yours to use as you see fit. See LICENSE if present.

Search Optimization (for GitHub & Google)

Short description (≤160 chars): Open-source toolkit to generate Claude Skills, Claude Code Agents, Slash Commands, and LLM Prompts—production-ready, validated, and scalable.

Primary keywords: Claude Skills, Claude Code Agents, Slash Commands, LLM Prompt Engineering, AI Agent Toolkit, Anthropic Claude, Production-Ready AI, MCP Tools. Secondary keywords: AI automation, developer toolkit, YAML frontmatter, typed Python, composable skills, prompt factory, code generation.

Version

Current: v1.3.0 • Updated: Oct 29, 2025

  • New: Master Slash Commands template; 17 presets; 3 official command patterns; strict 4-layer validation.
  • See full changelog in

Call to Action

Clone this wiki locally