# dsh-video-understand — DeepSeek Harness plugin install manifest

> Generated: 2026-09-17 21:28 · Source: https://deepseekmodel.com/plugins

## Overview

Low-cost video understanding tool: a video_understand tool turns a Bilibili link / BV / local video into an AVIS info layer (ASR + scene structure + object tracks + YOLO labels) and returns summary + Q&A. Question-driven dynamic routing across layers (L0 ASR / L1 object tracks / L2 key-frame VLM), semantic-layer reuse for repeat questions, per-question budget cap. Python engine: core layer needs faster-whisper / opencv / yt-dlp (~200-300MB); the optional semantic layer adds ~2GB of torch / transformers / ultralytics. A bundled doctor --fix sets up the venv and installs both.

## Plugin info

| Category | Maintainer | GitHub stars | License | Primary language | Last push |
| --- | --- | --- | --- | --- | --- |
| Tools & Capabilities | ilps2 | 8 | — | — | — |

## Install

```bash
dsh plugin --profile web add dsh-video-understand
```

## Steps

1. Install and start DeepSeek Harness (DSH, MIT-licensed open-source agent runtime) — `npx @deepseek-ai/dsh web`
2. Run the command above; the CLI resolves the plugin and verifies its source
3. Confirm with dsh plugins list; restart Harness if needed

> To pin a revision, append #commit to the repository spec.

## Audit before installing

Plugins run with your dsh process permissions and may execute code during installation. Read the repository source and license first, check for destructive commands or excessive access, and pin a commit hash for reproducible installs. This manifest is compiled by DeepseekModel from a public plugin catalog for reference only and does not vouch for third-party plugins.

---

- Repository: https://github.com/ilps2/dsh-video-understand
- Plugin page: https://deepseekmodel.com/plugin-detail?id=ilps2%2Fdsh-video-understand
- Source: https://deepseek-harness-plugin.com/zh-CN/plugins/ · Generated 2026-09-16