dsh-omni-modal v0.1.0

First public release.

Adds an Omni-modal settings section to DeepSeek Harness for configuring OpenAI-compatible image / video generation endpoints, with switchable config profiles.

• generate_image - synchronous text-to-image; saves to disk and returns a preview card
• generate_video - asynchronous submit; returns a task id immediately
• check_video - poll the task; downloads the finished clip on success

Install

dsh plugin --profile <profile> add github:yzhclcl/dsh-omni-modal

Then restart DSH. The section appears under Settings -> Omni-modal settings.