Alibaba's Qwen team published an open-source set of plugins that add image, video, audio and 3D handling to existing command-line coding agents, including Claude Code, Codex, Gemini CLI and Qwen Code. Eight capabilities are installed separately, covering local file reading, cloud vision and speech services, long-video question answering, media generation and control of running Blender and FreeCAD sessions. The cloud-based tools require paid API keys, while local image, video and document reading does not.
What changed
Command-line coding agents handled text and code, with image, video and audio work needing separate tooling.
What it unlocks
Asking a terminal coding agent to read a 4K screenshot, transcribe a meeting recording with speaker labels, answer questions about a two-hour video, or drive a running Blender or FreeCAD session.
- 8 capability plugins
- 22 Blender tools
- 14 FreeCAD tools
What you need to act on it
- the uv/uvx Python launcher
- a DashScope API key for the cloud media tools
- a Serper key for web and reverse-image search
- ffmpeg and other system tools
- WSL2 on Windows
Sources