RadTask vs AgentRQ
Both RadTask and AgentRQ are human-in-the-loop task managers for AI agents, accessible over MCP — you keep a human approving what AI does. If you're choosing between them, here's the honest core difference.
The short version: AgentRQ centers on the review queue — a place to approve and track what your AI agents are doing. RadTask centers on doing the task — a real autonomous agent researches, drafts, writes documents, and schedules in a sandbox, and a deterministic risk classifier holds only the irreversible steps (sending, spending, deleting, publishing) for your one-tap approval.
Where RadTask is distinct
- It executes the task, not just routes/queues it — and hands back deliverables you can download.
- A deterministic safety classifier decides reversible (runs automatically) vs. irreversible (held for approval), with the agent's plan shown.
- A published, machine-verifiable safety contract at /.well-known/agent-safety.json.
- An operator cockpit — projects, a board view, a calendar (the agent can schedule onto it, held for approval), and persistent receipts.
- Any AI connects via the MCP server and drops tasks straight into your cockpit.
When to pick which
If you mainly need a review/approval queue over agents you already run, AgentRQ is built around that. If you want an AI that actually does your task list — safely, with the irreversible steps gated — and a cockpit any AI can plug into, that's RadTask.
AgentRQ is an independent product; check agentrq.com for their current features. RadTask is in private beta — a tool with a safety design, not a guarantee. Connect your AI →