steve

Centralized repository for Claude Code agent configurations, hooks, commands, skills, and templates.

Project Status

Metric	Value
Components	377 total (137 agents, 97 commands, 57 skills, 59 hooks, 27 templates)
Test Suite	415 tests passing
Coverage	82.5%
Python	3.10+

Overview

This repository provides a well-organized collection of reusable components for Claude Code, including:

• Agents - Sub-agent configurations organized by domain and specialization
• Hooks - Claude hooks for analysis, guards, workflows, and lifecycle events
• Commands - Slash commands for various development workflows
• Skills - Reusable skills with bundled resources
• Rules - Language and format-specific coding rules
• Templates - Templates for creating new components

Map of Contents

Core Modules

Module	Description	Location
Agents	Sub-agent configurations by domain	steve/agents/
Commands	Slash commands for workflows	steve/commands/
Skills	Reusable domain knowledge bundles	steve/skills/
Hooks	Event-driven automation	steve/hooks/
Templates	Component scaffolds	steve/templates/
Rules	Language-specific coding rules	steve/rules/
Helpers	Python utility modules	steve/helpers/

Documentation

Document	Description
Getting Started	Quick start guide
Installation	Detailed installation
Architecture	System design
Development	Development workflow
Using Agents	Agent usage guide
Using Commands	Command usage guide
Using Skills	Skill usage guide
Using Hooks	Hook usage guide
Contributing	Contribution guidelines
Scripts Reference	Script documentation
FAQ	Frequently asked questions

Repository Structure

steve/
├── agents/              # Sub-agent configurations
│   ├── core/           # Core system agents
│   ├── development/    # Development workflow agents
│   ├── code-quality/   # Code quality and review agents
│   ├── expert-advisors/# Domain expert advisors
│   └── ...             # More specialized domains
│
├── hooks/               # Claude hooks
│   ├── analyzers/      # Code analysis hooks
│   ├── guards/         # Guard hooks (pre-commit checks)
│   ├── workflows/      # Workflow automation hooks
│   ├── lifecycle/      # Lifecycle event hooks
│   └── context/        # Context management hooks
│
├── commands/            # Slash commands
│   ├── _team/          # Team collaboration commands
│   ├── agents/         # Agent management commands
│   ├── git/            # Git workflow commands
│   ├── dev/            # Development commands
│   └── ...             # More command categories
│
├── skills/              # Reusable skills
│   ├── skill-name/     # Each skill in its own directory
│   │   ├── SKILL.md    # Main skill definition
│   │   ├── references/ # Documentation references
│   │   ├── scripts/    # Executable code
│   │   └── assets/     # Output files
│   └── ...
│
├── rules/               # Language-specific rules
│   ├── python.md
│   ├── typescript.md
│   ├── shell.md
│   └── ...
│
├── templates/           # Component templates
│   ├── AGENT_PLAYBOOK.template.md
│   ├── AGENT_SKILL.template.md
│   ├── SLASH_COMMAND.template.md
│   └── ...
│
├── helpers/             # Python helper modules
│   ├── context_monitor.py
│   ├── debug_rotation.py
│   ├── history_archival.py
│   └── ...
│
└── scripts/             # Python management scripts
    ├── build_index.py
    ├── install_component.py
    ├── list_components.py
    └── ...

Component Organization

Domain/Specialization Hierarchy

Components are organized by domain and specialization to enable:

• Easy discovery - Find components by domain
• Logical grouping - Related components are co-located
• Scalability - Add new domains without restructuring

Naming Conventions

• Agents: kebab-case (e.g., code-reviewer, dependency-manager)
• Skills: kebab-case (e.g., code-context-finder, tdd-pytest)
• Commands: kebab-case.md (e.g., create-subagent.md)
• Hooks: kebab-case.py and kebab-case.md pairs

Quick Start

Installation

# Clone the repository
git clone https://github.com/joe/steve.git
cd steve

# Install dependencies
uv sync

# Build the component index
uv run scripts/build_index.py

# Run tests
uv run pytest

# Index
make run-index                              # Build index
make run-index ARGS="--verbose"             # With options

# List
make run-list                               # List all
make run-list ARGS="--type agent"           # Filter by type

# Install
make run-install ARGS="python-pro"          # Install component

# Publish
make run-publish                            # Publish ALL
make run-publish ARGS="steve/agents/core/*.md"  # Glob
make run-publish ARGS="steve/agents/core/example-agent.md"  # Single

# Secrets
make run-secrets ARGS="--scan"              # Scan
make run-secrets ARGS="--baseline"          # Baseline

# Metadata
make run-metadata ARGS="steve/agents/core/example-agent.md"
make run-metadata ARGS="steve/agents/core/*.md"  # Glob
make run-metadata ARGS="steve/agents/core/*.md --gist-url https://..."  # Glob with options

# Batch metadata
make run-batch-metadata ARGS='steve/agents --dry-run'
make run-batch-metadata ARGS='steve/agents --key version 1.0.1'
make run-batch-metadata ARGS='steve/agents --key author "Joseph OBrien" --key status unpublished --key updated 2025-12-23 --key version 1.0.1 --dry-run'
make run-batch-metadata ARGS='steve/agents/core --pattern "*.md" --key category core'
make run-batch-metadata ARGS='steve/skills --pattern "**/SKILL.md" --key type skill'

Using Components

# List available components
uv run scripts/list_components.py

# Search for specific components
uv run scripts/list_components.py --search "code review"

# Install a component
uv run scripts/install_component.py code-reviewer

Creating New Components

1. Use templates from steve/templates/
2. Follow naming conventions (kebab-case)
3. Include YAML frontmatter where applicable
4. Add README.md for complex components

File Formats

Agent Files

---
name: agent-name
description: Action-oriented description
tools: Read, Write, Grep
model: sonnet
color: cyan
skills: skill1, skill2
---
# Purpose
---
## Instructions

Skill Files

---
name: skill-name
description: Third-person description
---
# Skill Title

Contributing

1. Follow the domain/specialization structure
2. Use kebab-case naming conventions
3. Include YAML frontmatter where applicable
4. Add README.md files for complex components
5. Run secrets detection before committing (see Security section)

Security

• Secrets Detection: Configured via .pre-commit-config.yaml
• Sensitive Files: Excluded via .gitignore
• Public Repository: Ensure no secrets are committed

Scripts Reference

Script	Description
build_index.py	Build component index from repository
list_components.py	List and search components
install_component.py	Install components to Claude Code
install_from_gist.py	Install from GitHub Gist
publish_to_gist.py	Publish components to Gist
publish_registry.py	Publish registry to Gist
add_metadata.py	Update component frontmatter
detect_secrets.py	Scan for secrets in codebase
python_to_markdown.py	Convert Python to Markdown

Documentation

• Getting Started
• Architecture
• Development Guide
• Contributing
• Scripts Reference
• FAQ

See CLAUDE.md for Claude Code-specific configuration and docs/ for complete documentation.