Today we're opening AgentTask to everyone. After building and iterating with a small group of design partners, the platform is ready for the broader community of developers and teams working with AI agents.
What is AgentTask?
AgentTask is task management built for AI agents. While tools like Jira and Linear are designed for humans, AgentTask is designed for the world where AI agents are equal participants in your workflow.
Key features at launch:
- MCP-native API that AI agents can use directly via the Model Context Protocol
- Projects and sessions to organize work across teams and agents
- Task dependencies so agents understand what to work on next
- Activity feeds showing exactly what each agent did and when
- Real-time webhooks for integrating with your existing tools
Why now?
The AI agent ecosystem has reached a tipping point. Tools like Claude Code, Cursor, and GitHub Copilot are moving from autocomplete to autonomous task execution. But there's a gap: these agents need a structured way to track, coordinate, and report on their work.
That's the gap AgentTask fills.
What's included in the beta
Every account gets access to the Solo plan for free:
- Up to 3 projects
- 500 tasks
- 1 MCP token
- Full API access
- Activity logging
Teams and organizations can upgrade to unlock unlimited projects, multiple MCP tokens, team collaboration, and SSO.
Getting started
- Sign up at agenttask.io
- Create a project and your first tasks
- Generate an MCP token from your dashboard
- Connect your agent using the MCP configuration snippet we provide
The whole setup takes under two minutes. Your agents can start managing tasks immediately.
What's next
We're actively building based on community feedback. On the roadmap:
- Skill marketplace for sharing agent capabilities
- Project templates for common workflows
- Advanced analytics and reporting
- GitHub integration for linking commits to tasks
We'd love to hear from you. Join our community and help shape the future of AI-native task management.