What should an AI-agent file workspace do next?

We’re building AgentDrive, a persistent file and workspace layer for AI agents accessed primarily through MCP.

Today, agents can create and organize files and folders, track revisions, upload larger files, restore work, and share selected resources. We’re opening the private beta and want to learn from people building coding agents and agent infrastructure.

A few questions:

1. What’s the first workflow you’d ask an agent to perform with a persistent workspace?

2. Which storage tools or primitives are missing from the current MCP surface?

3. What would make you trust an agent-storage product with important work?

4. Would you prefer a hosted MCP service, a future REST/SDK integration, self-hosting, or some combination?

Details: https://tokencanopy.com/products/agentdrive/

Hosted MCP: https://drive.mcp.tokencanopy.com/mcp

On the roadmap:

• API/SDK with API support

• Open source

• Support more file formats

• Mobile and Desktop support

Your feedback is important to help us understand what to prioritize and build next!

Feel free to reach out at support@team.tokencanopy.com or to me directly at josh@tokencanopy.com