Skip to content

How to Use the Xiaomo Assistant (Web)

‍

Xiaomo (小莫) is the lightweight assistant built into Mobius — it floats at the bottom-right of every page. Open it to ask, follow up, spawn clones, attach files, or go hands-free with voice — no need to create a project or task first. This walkthrough covers the whole flow.

1. Open Xiaomo

  • Every page has a round "Xiaomo" button at the bottom-right. Click it to open the assistant panel.

image

2. The Panel Layout

  • It opens as a floating layer: the session list on the left (your main Xiaomo + any clones), the conversation in the middle, and the input box at the bottom.

image

3. Ask, Then Wait for the Reply

  • Type your question in the input box and press Enter or click "Send". Your first question auto-creates a "Main Xiaomo" session; later follow-ups reuse it (the context is kept).

image

  • Replies support Markdown — code blocks, lists and tables are all rendered. For long replies, click "Expand" at the top to switch to a larger / fullscreen view (the screenshot below shows the expanded view).

image

4. Top Toolbar

  • A row of small buttons across the top of the panel. The ones you'll use most:

image

  • Open in new window: pop the current session into its own tab so you can keep it open.
  • Skills: view / edit the skills (Skills) Xiaomo uses, inside VSCode.
  • Preset: predefine a Xiaomo session template to reuse later (see section 7).
  • Expand / Fullscreen: compact → expanded → fullscreen, cycling through three sizes.
  • Close: collapse the panel (the conversation is preserved).

5. Spawn a "Clone Xiaomo" to Run a Task in Parallel

  • Want Xiaomo to do another thing at the same time without interrupting the current chat? Click "Spawn clone", describe the single task this clone should handle, pick a model, and start it.

image

  • A clone handles only that one task, then reports the result back to the main Xiaomo. A new "Clone Xiaomo #N" appears in the session list on the left.

6. Voice & Attachments

  • A row of small buttons on the right of the input box:

image

  • Upload image / file: click to choose, or paste a screenshot / drag a file straight into the input box.
  • Voice input: click once, speak, and your words are transcribed and sent.
  • Expand input: enlarge the box for longer messages.
  • At the top there's also a "Enable reply voice playback" toggle — turn it on and Xiaomo will read each reply aloud (voice is selectable).

7. Preset (Advanced, Skippable)

  • Click "Preset" to predefine a Xiaomo session template: first fill in the session name and purpose / problem to solve, then pick model and skills in the next step. Save it and reuse it next time instead of re-describing.

image

‍

In one sentence: Xiaomo is an always-on helper — open it anytime, ask anytime, and spawn a clone when you need to parallelize.