WebUI
The WebUI is the browser app, the same on freegent.ai and on a local server. The left rail has New Chat, Project, Tasks, Chats, a dark-mode toggle and Settings.
| Page | What it is for |
|---|---|
| Project | The agent's workspace: file tree, editor, import and export, previews of images, PDFs, Office files and HTML |
| Tasks | The task board and Autopilot |
| Chats | The conversation, saved chats, attachments, voice, and the Chat and Cowork modes |
| Settings | Every Settings tab, profiles, and where your data lives |
First chat
- Open Settings → Models → API Credentials and paste a key for one provider (see Default providers).
- Type a request and press Enter. Shift+Enter adds a newline and Esc stops generation.
- The agent may call tools: read and write project files, run code, search the web. Each call appears as a step in the chat. Files it creates land in Project.
The toolbar shows the active model and how much of its context window is used (4.2k / 128k, yellow at 70%, red at 90%). FreeGent compacts old history automatically; see Context and compaction.
Where things run
The page itself is static. Provider calls go through a proxy (How requests reach providers) and code runs in browser sandboxes (Code execution). On freegent.ai the agent cannot run bash or Python on your machine, use run_git or reach models on localhost; the local server adds those.