Skip to content

🎉 Rigger v1.0.0 - Initial Release

Choose a tag to compare

@BitBrujo BitBrujo released this 24 Nov 03:04
· 21 commits to main since this release

Rigger v1.0.0 - Initial Production Release

Rigger is a comprehensive visual testing interface for the Claude Agent SDK. This initial release provides a full-featured web application for configuring AI agents, testing tools, monitoring execution, and managing agent workflows.


🚀 Feature Overview

Core Agent SDK Integration

  • 19 Built-in Tools with categorized organization:
    • 📁 File Operations: Read, Write, Edit, Glob, Grep
    • ⚙️ Execution: Bash, BashOutput, KillShell
    • 🌐 Web Access: WebFetch, WebSearch
    • 📋 Planning & Interaction: TodoWrite, Task, AskUserQuestion, ExitPlanMode
    • 🤖 Agent System: Skill, SlashCommand
  • Real-time Streaming: Server-Sent Events (SSE) for responsive UI updates
  • Debug Metrics: Comprehensive tracking of tokens, costs, cache statistics, and timing
  • Conversation Persistence: Full chat history with database storage

Session Management

  • Persistent Execution Contexts: Track agent activity across multiple requests
  • Two-Tier Emergency Stop:
    • ⏸️ Graceful Stop: Allow current operation to finish cleanly (5-second grace period)
    • 🛑 Force Kill: Immediate termination with complete resource cleanup
  • Resource Monitoring: Real-time tracking of tokens, costs, turns, and tool usage
  • Auto-Cleanup: Automatic termination of idle sessions (5-minute timeout)

Configuration Management

  • Presets System: Save and load complete agent configurations as named presets
  • Import/Export: JSON-based configuration portability for sharing
  • Change Detection: Visual indicators when active config differs from loaded preset
  • 30+ Configuration Parameters: Full control over Agent SDK settings

Advanced Capabilities

🔌 MCP (Model Context Protocol) Servers

External plugin support for extending Claude with additional tools:

  • Pre-configured templates for: GitHub, Playwright, Notion, Filesystem, Git, Memory, Time
  • Custom environment variable configuration
  • Dynamic tool discovery

📚 Skills System

  • Packaged agent workflows loaded from .claude/skills/
  • Built-in examples: PDF Processing, Data Transformation, Code Review
  • Auto-discovery and management UI
  • Create, edit, and delete skills directly from the interface

🤝 Subagents

Specialized AI agents with custom prompts and tools:

  • Built-in templates: Code Reviewer, Bug Hunter, Doc Writer, Refactorer, Test Generator
  • Independent model and temperature settings
  • Tool access restrictions for security

🪝 Hooks System

Event-driven automation with pre-built templates:

  • Triggers: on-prompt-submit, on-response-complete, on-tool-use, on-error
  • Templates: Git Auto-Commit, Slack Notifications, Error Logging, Code Formatting, Backups, API Webhooks
  • Variable interpolation for dynamic configurations

📎 File Upload System

  • Upload files up to 10MB for agent context
  • Supported formats: text, JSON, CSV, PDF, images, code files
  • Multiple integration methods: System Prompt, Working Directory, or Both
  • Global vs conversation-scoped files

🖥️ User Interface

Three-Panel Layout

The application features a responsive three-panel design:

Left Panel - Sidebar Navigation (10 tabs):

  • Sessions, Configuration, Basic Config, Tools
  • MCP Servers, Skills, Subagents, Hooks, Files, Advanced

Center Panel - Configuration Content:

  • Tab-specific configuration and management UIs
  • Real-time form validation
  • Persistent settings across sessions

Right Panel - Chat Interface (4 tabs):

  • 💬 Chat: Message history with streaming responses
  • 🐛 Debug: Token usage, costs, cache metrics, API latency
  • 🔧 Tools: Tool execution timeline with parameters and results
  • ✅ Todo: Agent-created task list visualization

🛠️ Technical Architecture

Frontend Stack

  • Framework: Next.js 16 with React 19
  • Language: TypeScript (full type safety)
  • Styling: Tailwind CSS 4 with shadcn/ui components
  • State Management: Zustand (no provider wrapping needed)
  • Real-time: Server-Sent Events (SSE)

Backend Stack

  • Framework: Express.js with Node.js
  • Language: TypeScript
  • Database: PostgreSQL with JSONB for flexible schema
  • Agent SDK: @anthropic-ai/claude-agent-sdk with full tool support
  • Containerization: Docker Compose for easy deployment

Database Schema

8 core tables with optimized indexing:

  • conversations: Chat history and messages
  • agent_sessions: Session lifecycle, metrics, emergency control
  • presets: Saved agent configurations
  • usage_logs: Per-request metrics and analytics
  • tool_usage_logs: Per-tool execution tracking
  • todos / todo_items: Task list management
  • custom_agents: User-defined subagent definitions
  • uploaded_files: File uploads for agent context

Key Design Decisions

  • Session-based execution: Persistent contexts with abort controls
  • Event-driven hooks: Automation triggers for common workflows
  • Type-safe throughout: Full TypeScript coverage for reliability
  • Docker-ready: Complete containerization with single-command startup
  • Real-time streaming: SSE for responsive UI updates without polling

📦 Installation Guide

Prerequisites

Quick Start

  1. Clone the repository:

    git clone https://github.com/BitBrujo/rigger.git
    cd rigger
  2. Configure environment variables:

    Create backend/.env:

    DATABASE_URL=postgresql://agent_user:agent_pass@postgres:5432/agent_db
    ANTHROPIC_API_KEY=your_api_key_here
    PORT=3001

    Create .env.local:

    NEXT_PUBLIC_API_URL=http://localhost:3333/api
  3. Start backend and database:

    docker-compose up -d

    Wait ~10 seconds for PostgreSQL initialization.

  4. Install frontend dependencies:

    npm install
  5. Start development server:

    npm run dev
  6. Access the application:

    Open http://localhost:3334 in your browser.

Port Configuration

  • Frontend: 3334 (Next.js)
  • Backend API: 3333 (Express - external)
  • Database: 5335 (PostgreSQL - external)

Verification

Test the backend health endpoint:

curl http://localhost:3333/health

📚 Documentation

  • README.md: Project overview and quick start guide
  • CHANGELOG.md: Complete feature documentation and version history
  • CLAUDE.md: Comprehensive architecture and developer documentation (1000+ lines)
    • Complete application structure
    • Request flow diagrams
    • Database schema details
    • Development patterns
    • Troubleshooting guide
  • Skills Documentation: .claude/skills/README.md

🎯 Use Cases

Development & Testing:

  • Test Claude Agent SDK integrations before production deployment
  • Experiment with different tool combinations and configurations
  • Debug agent behavior with comprehensive metrics

Workflow Automation:

  • Create custom skills for repetitive tasks
  • Set up hooks for event-driven automation
  • Integrate external tools via MCP servers

Team Collaboration:

  • Share agent configurations via presets
  • Export/import JSON configurations across projects
  • Manage conversation history for review

Learning & Experimentation:

  • Explore Claude's capabilities with 19 built-in tools
  • Understand token usage and costs with real-time metrics
  • Test subagents with specialized prompts

📸 Screenshots

The application features a modern, responsive interface with:

  • Dark mode support with next-themes
  • Resizable panels for customized layouts
  • Real-time streaming with visual indicators
  • Comprehensive debug panels showing all API metrics
  • Tool execution timeline with parameters and results
  • Session controls with emergency stop buttons

(Screenshots can be added to the repository wiki or documentation site)


🔮 Future Plans

  • Enhanced analytics dashboard with charts and visualizations
  • Team collaboration features (shared sessions, comments)
  • Custom theme support beyond dark/light modes
  • Plugin marketplace for MCP servers and skills
  • API documentation generator from tool usage
  • Performance profiling tools for optimization
  • Export conversations to various formats (PDF, Markdown, JSON)

🤝 Contributing

Contributions are welcome! Please see the repository for contribution guidelines.


📄 License

This project is licensed under the MIT License. See the LICENSE file for details.


🙏 Acknowledgments

Built with:


Thank you for trying Rigger v1.0.0! 🎉

For issues, questions, or feature requests, please visit the GitHub Issues page.