.agents

Agents, skills, commands, and rules for AI-powered coding tools — used daily to supercharge development productivity.

This repository contains reusable agent personas, workflow commands, specialized skills, coding rules, and system hooks that work with any AI coding assistant that supports markdown-based configuration.

📦 What's Inside

.agents/
├── agents/          # 7 specialized AI agent personas
├── commands/        # 7 workflow automation commands
├── skills/          # 3 domain-specific skills
├── rules/           # 4 coding convention rules
└── claude-code-hooks.json   # Hook configuration (Claude Code specific)

🚀 Quick Start

Option 1: Clone Entire Setup

cd ~
git clone git@github.com:Weaverse/.agents.git

# Symlink into your AI tool's config directory
# For Claude Code:
ln -s ~/.agents/agents ~/.claude/agents
ln -s ~/.agents/commands ~/.claude/commands
ln -s ~/.agents/skills ~/.claude/skills

Option 2: Copy Individual Items

1. Browse the agents/, commands/, skills/, or rules/ folders
2. Copy the files you need into your AI tool's config directory

🤖 Available Agents

Production-tested agent personas used daily by the Weaverse development team:

Agent	Purpose	Best For
code-review-specialist	Expert code reviews with security & performance analysis	PR reviews, quality assessment
debug-specialist	Systematic debugging and issue diagnosis	Runtime errors, test failures, build issues
solution-architect	Strategic guidance with pros/cons analysis	Architecture decisions, complex problems
docs-writer	Clear, simple technical documentation	READMEs, API docs, guides
code-explainer	Break down complex code concepts	Code walkthroughs, onboarding
feature-architecture-reviewer	Review feature implementations	Scalability assessment, design validation
pr-review-analyst	Deep PR analysis with impact assessment	Change reviews, regression prevention

View detailed agent descriptions →

⚡ Workflow Commands

Automation commands for common development tasks:

• create-pr - Create pull requests with AI-generated descriptions
• fix-pr - Fix issues from PR reviews
• format - Auto-format code and fix TypeScript issues
• review-changes - Review uncommitted changes
• review-code - In-depth code quality analysis
• handoff - Generate context for team handoffs
• ultrathink - Deep reasoning for complex problems

🎯 Specialized Skills

Domain-specific procedures for complex tasks:

• code-review - Comprehensive code audits (dead code, duplications, patterns, bugs)
• feature-plan - Detailed implementation planning before coding
• frontend-design - Create distinctive, production-grade UI components

📏 Coding Rules

Opinionated conventions enforced across the team:

• coding - Naming, styling, and general coding standards
• commit - Commit message conventions
• never-auto-commit - Prevent AI tools from auto-committing
• spec-driven-development - Write specs before code

🔔 System Hooks

The included claude-code-hooks.json and claude-code-notifier.sh enable system notifications for:

• Session start/end
• Task completion
• Custom alerts

Cross-platform support: macOS, Linux, Windows (PowerShell).

Note: Hooks are currently Claude Code specific. Contributions for other tools welcome.

🛠️ Development

See AGENTS.md for:

• Validation commands
• Code style guidelines
• Frontmatter requirements
• Quality checklist

🤝 Contributing

Contributions welcome! When adding new agents/commands/skills:

1. Follow the naming convention: kebab-case.md
2. Include proper YAML frontmatter
3. Add 2-3 realistic examples with <commentary> tags
4. Test manually in your AI coding tool
5. Update relevant README sections

📝 License

MIT

---

Built by Weaverse — Empowering developers with AI-enhanced workflows.