Praktor

Praktor is an innovative personal AI agent assistant designed for self-hosting, providing a robust alternative to existing solutions like OpenClaw. With a primary focus on integrating with Telegram and utilizing Claude Code, Praktor allows users to define distinct agents with specific roles and messaging capabilities directly through Telegram, while managing all activities via a centralized Mission Control user interface.

Architecture Overview

Praktor employs a streamlined architecture that facilitates effective interaction between components:

Telegram ──> Go Gateway ──> Router ──> Embedded NATS ──> Agent Containers (Docker)
                 |                                            |
             SQLite DB                                Claude Code SDK
                 |
       Mission Control (React SPA)

The entire system operates through a single Go binary that seamlessly orchestrates the message flow, ensuring agents can be executed in isolated Docker containers while providing real-time responses back to users.

Key Features

• Mission Control: A real-time dashboard equipped with WebSocket updates to monitor all agent activities.
• Telegram Integration: Communicate with agents via Telegram, utilizing commands such as /start, /stop, and /reset.
• Agent Management: Create multiple named agents, each configured with unique models and roles.
• Smart Routing: Direct messages to specific agents using an @agent_name prefix or AI-driven default routing.
• Isolated Agent Environments: Each agent operates within its designated Docker container, enhancing security and management.
• Persistent Storage: Employ SQLite databases for maintaining memory across sessions, allowing agents to recall facts and user profiles.
• Configuration Management: Change configurations live with hot reload capabilities, ensuring seamless updates without downtime.
• Scheduled Task Management: Execute cron jobs, intervals, or one-shot tasks through agents with results communicated via Telegram.
• Secure Vault: Store sensitive credentials and tokens securely, enhancing the application's security model.
• Agent Swarm: Manage complex tasks through the orchestration of multiple agents, utilizing fan-out and collaborative patterns.
• Backup and Restore: Facilitate full backup and restore operations for all operational data, ensuring system integrity and recovery options.

Use Cases

Praktor is ideal for users seeking to automate tasks through AI, whether it be engaging in conversation, conducting web research, or performing scripted tasks like browser automation. Interaction commands enable users to direct agents effectively, providing flexibility and ease of use:

Hello!                               → routed to default agent  
@coder fix the login bug             → routed to the coder agent  
@researcher find papers on RAG       → routed to the researcher agent

Browser Automation

Praktor integrates the playwright-cli for browser automation capabilities. Agents can execute actions such as navigating websites and filling forms without needing extensive configuration.

Management and Configuration

The project's configuration is managed via a simple YAML file, with clear documentation on editing agent definitions, Telegram settings, and enabling additional features such as Nix package management for on-demand tool installation.

Praktor provides extensive documentation to assist users in deploying, configuring, and expanding their AI agent capabilities, promoting a highly customizable AI environment.

For developers, tools for building and testing are available, ensuring contributors can easily integrate with the project's ongoing development cycle.

In summary, Praktor combines the power of personal AI agents with user-friendly management, real-time interaction, and highly customizable configurations, making it a powerful tool for individuals and teams looking to enhance productivity through automation.