AgentRQ

AgentRQ lets people assign work to AI agents, follow their progress and answer requests for decisions from a browser, phone or desktop app. Agents run on computers you own; the task dashboard can run on your server or the hosted service.

• Tasks and approvals: Each agent has a workspace with its own queue and conversations, shown in a shared list or Kanban board where you can reply and approve actions.
• Run inspection: Search tool-call histories to see inputs, results and approval decisions, or open a running agent's terminal remotely.
• Scheduling and workflows: Schedule one-time or recurring tasks and connect completion events to other workspaces through a visual workflow graph.
• Persistent context: Workspace notes and shared skills carry instructions between tasks and machines; workspace forks give parallel agents separate queues and working folders while sharing memory and skills.
• Local input assistance: An in-browser LLM generates task titles, and desktop voice transcription uses local Whisper; mobile voice input uses the phone's dictation.

AgentRQ connects through MCP and supports Claude Code and DeepSeek Harness natively. Its ACP gateway connects compatible agents including Gemini CLI, Codex CLI, Cursor, GitHub Copilot and OpenCode. Slack integration provides a shared channel per workspace with attributed approvals.

The project is open source under AGPL-3.0. Self-hosting keeps the task service on your server, while the hosted offering runs that service externally. Running an agent on your computer doesn't make its model local; model processing depends on the connected agent and its provider.