DSH media panel with copyable image answers and local video preprocessing. Vision inference uses the selected DSH model.
DSH integration
Native runtime
Author-claimed
Safety audit
Unaudited
Last verified
2026-10-19
License
MIT
01What can it help you accomplish?
Understand images and video within DSH.
Model answers, copyable image history, normalized videos and keyframes.
DSH users working with image and video attachments.
02How to install into DeepSeek Harness
Prerequisites
- Plugin 0.4.x targets DSH 0.2.0-rc.2; keep 0.3.2 for DSH 0.1.x.
- For video, install FFmpeg/FFprobe >=6.1 with matching major versions and libx264, plus PySceneDetect >=0.7.1 <0.8. All commands must be on the PATH of the account/environment running DSH.
Installation steps
- 01
For video on macOS, run `brew install ffmpeg uv`; on Arch, `sudo pacman -Syu ffmpeg uv`; on Windows x64, `winget install --id Gyan.FFmpeg --exact` and `winget install --id astral-sh.uv --exact`.
$ brew install ffmpeg uv
- 02
For video, run `uv tool install --python 3.12 "scenedetect>=0.7.1,<0.8"` and `uv tool update-shell`, then reopen the terminal.
- 03
Install with `dsh plugin --profile web add dsh-visual-plugin`. Restart dsh web from the environment where the dependency checks passed, and select an image-capable model in DSH.
$ dsh plugin --profile web add dsh-visual-plugin
Verify the integration
- For video, check `ffmpeg -version`, `ffprobe -version`, `ffmpeg -hide_banner -encoders`, and `scenedetect version` for the required versions and libx264.
- Send an image: the panel records its thumbnail and the current model’s final answer.
Rollback
- Run `dsh plugin --profile web remove dsh-visual-plugin`, then restart dsh web.
03DSH integration and capability boundaries
Native DSH image history, video tools and right-side media panel.
Native image history
Uploaded image→Current model answer stored alongside a thumbnail for copying
Video preprocessing
Uploaded video→Normalize video to H.264/yuv420p MP4 and extract timestamped keyframes for the current DSH vision model.
Video preprocessing and plugin-owned storage remain on the DSH machine, but images and keyframes are sent to the selected model through DSH. Local inference depends on the DSH model configuration.
04Who is it for? When not to use it?
Good for
- DSH users working with image and video attachments.
Not for
- For video, install FFmpeg/FFprobe >=6.1 with matching major versions and libx264, plus PySceneDetect >=0.7.1 <0.8. All commands must be on the PATH of the account/environment running DSH.
- Video preprocessing and plugin-owned storage remain on the DSH machine, but images and keyframes are sent to the selected model through DSH. Local inference depends on the DSH model configuration.
05Compatibility, maintenance and safety notes
- For video, install FFmpeg/FFprobe >=6.1 with matching major versions and libx264, plus PySceneDetect >=0.7.1 <0.8. All commands must be on the PATH of the account/environment running DSH.
- Video preprocessing and plugin-owned storage remain on the DSH machine, but images and keyframes are sent to the selected model through DSH. Local inference depends on the DSH model configuration.
README claims pinned to a commit; runtime not tested.
06Frequently asked questions
Which DSH version does 0.4.x target?
Plugin 0.4.x targets DSH 0.2.0-rc.2; keep 0.3.2 for DSH 0.1.x.
Do image features need FFmpeg?
No. Image features work without the video tools; video requires the documented FFmpeg/FFprobe and PySceneDetect setup.
Does all inference stay local?
Video preprocessing and plugin-owned storage remain on the DSH machine, but images and keyframes are sent to the selected model through DSH. Local inference depends on the DSH model configuration.
How do I uninstall from the DSH web profile?
Run `dsh plugin --profile web remove dsh-visual-plugin`, then restart dsh web.
07Related DSH workflows
dsh-image-gen
by shanliuling
Adds in-chat multimodal image generation via a `generate_image` tool with multi-provider support (Gemini/OpenAI/Seedream), interactive preview, copy/download, and session-attached results.
dsh-vision-router
by ysr666
Eyes for text-only DeepSeek Harness agents: built-in free vision chain (no key) + pixel-level vision tools (Q&A, grounding, crop, pixel diff, colors, OCR, SVG trace, cutout, screenshots). One-command install, no Python, image turns work like ordinary tool-calling turns.
dsh-vision-toolkit
by anionex
为纯文本 DSH Agent 提供 10 个结构化视觉工具:意图感知图片问答、长截图 OCR、原始像素 grounding、UI 还原、像素 diff 等
dsh-see-now
by saos-3624
Captures screen, OBS, or webcam frames, filters by visual change, packs them into image sheets, and delivers them to DSH sessions.
08Data and sources
A local-first visual media plugin for DeepSeek Harness.
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-10-19. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.
