Documentation
How agentleFS works, how to connect an agent over MCP, and how to call the REST API from your own code. No account needed to read any of it.
What is agentleFS?
agentleFS is the control plane for the content your AI agents read. You organize knowledge into versioned folders, decide exactly what each agent is allowed to see with role-based access, and keep a full audit trail of every read and write. This console is where operators upload content, approve what goes live, and manage access. People connect their agent clients by signing in through the browser; agents your application runs authenticate with a scoped token.
Getting started
- 1. Create a top-level folder
Files → New top-level folder. Name it and add your first files. The folder is created on upload.
- 2. Set access
In Permissions, grant a person or a group read, write or approver (can share and manage access, and edit) on a folder or a single file. A grant on a folder reaches everything inside it.
- 3. Connect your agent
Point your client at the /mcp URL on the Connect tab and sign in through the browser. No token to mint or paste.
- 4. Mint tokens only for agents
Agents your application runs — in-app assistants, scheduled jobs, CI — have no browser, so give each its own principal and token.