Hi everyone. I built Stuga because I wanted AI agents writing into our team's documents, but not silently. Claude Code and Codex can already edit files. In a shared document that's unnerving: you find out afterwards, if at all. In Stuga, agents' document-text edits arrive as suggestions, marked up like tracked changes in Word or Google Docs, and you accept or reject each one in the document. Edits people make go straight in. What's in it * Documents written together in real time, with comments and version history * Databases with read-only SQL, saved views and a page for every row * MCP built in: the Claude desktop app, Claude Code and Codex connect on the plan you already pay for * Review AI edits: one inbox for every agent run, showing which agent made it and what it changed * Ask: answers from your own documents and tables, with sources * Search that never shows a document you can't open Where it runs: a machine you own. A Mac with an M1 chip or newer on macOS 13+, or a Linux server or NAS with Docker. There's no telemetry. Your teammates just open it in a browser. Free and open source. Waiting for review is the default. A document's owner can let agents apply changes at once; those changes are still recorded and can be reverted where possible. I'd love to hear where you'd let an agent write freely, and where you'd want to review first.

Stuga

Tracked changes for AI agents, on a machine you own

Tracked changes for AI agents, on a machine you own