dorkos

DorkOS serves as a centralized 'mission control' interface for managing multiple AI coding agents, specifically supporting tools such as Claude Code, Codex, and OpenCode. Designed for developers who run agents from various sources (terminals, editors, scripts), it aggregates all sessions into a single browser-based dashboard. Key capabilities include: - **Session Monitoring**: View the status, history, and rich markdown output of every agent session in one place, regardless of where it was initiated. - **Action Approval**: Review and approve or deny specific actions taken by agents, providing a safety layer for autonomous operations. - **Task Scheduling**: Define and schedule agent runs (e.g., nightly audits, morning builds) without keeping terminals open. The system handles duplicate prevention and persistent scheduling. - **Inter-Agent Communication (Relay & Mesh)**: Agents can notify users via Telegram, webhooks, or the browser. They can also discover and communicate with each other across different projects using a mesh network approach. - **MCP Integration**: Supports the Model Context Protocol (MCP), allowing other AI tools to interact with DorkOS's local environment and tools securely. The software is self-hosted, MIT-licensed, and runs locally on the user's machine, ensuring data privacy. It requires Node.js 22.22.3+ and can be installed via npm or Docker. Installation leverages existing authentication for supported agents like Claude Code, or accepts API keys directly.