
Features
- Delivery checks for tasks. Add checks to a task and a Delivery Checker reviews the result before it's delivered. A failed check can hold the result back and start a repair round, and you get a verification report with the verdict, score, and the evidence behind each check.
- Send audio to an agent. Attach an audio file such as an MP3, and audio-capable models like Gemini can listen to it. Audio messages now play inline with a waveform player.
- Turn a document into a skill. A new "Convert to skill" action in the document explorer turns any agent document into a reusable skill in place.
- Self-update the CLI. Run
lh updateto upgrade the LobeHub CLI to the latest version, and usefile uploadto send local files.
Improvements
- Creating an agent now recommends relevant skills from the Skill Store.
- GLM-5.2 lets you choose how much it thinks before answering.
- Built-in Linear results render as cards instead of raw JSON.
- A cold start shows one continuous loading screen instead of a brief logo flash.
- Topics holding unsent text show a "[Draft]" hint, and unread status now syncs from the server so it stays consistent across devices.
- Desktop: jump to a tab with Mod+1–9, cycle tabs with Ctrl+Tab, and read update notes as formatted Markdown.
- The execution device switcher gained a Manage button, and the device detail panel lists recent directories.
- When an agent fails in a connected chat app, the reply explains why instead of showing a bare operation ID.
Fixes
- Editing a connector no longer silently wipes its saved credentials on save.
- Switching the interface language no longer flashes a blank screen.
- Adding a custom model with a duplicate ID is now rejected up front.
- The agent working-directory clear button now actually clears it.
- Search results no longer fail on a citation with an empty link.
- WeChat and QQ file attachments, including audio, are recovered more reliably.
