dsh-video-lens

Video understanding for DeepSeek Harness — give text-only agents eyes and ears on video.

A DeepSeek Harness (DSH) plugin that lets text-only LLM agents understand local video files. It provides two tools:

Tool	What it does
video_probe	Cheap, instant metadata via ffprobe: container, duration, resolution, fps, codecs, audio tracks, subtitles.
video_analyze	Content understanding: scene-change-aware frame sampling (ffmpeg scdet), optional ASR transcript (speech with timestamps), fused with any OpenAI-compatible vision model into structured evidence JSON.
video_ask	Time-anchored Q&A: parses explicit time references ("at 3:20", "第2分钟") or locates relevant speech via transcript keyword matching, re-samples frames from the matched windows, and answers with grounded evidence (answer + confidence + supporting timestamps).

v0.3.2. The plugin never locks you into a provider: vision and ASR are both OpenAI-compatible endpoints configured via baseUrl + model + key env var.

How it works

video file ──► video_probe ──► ffprobe ──► compact metadata JSON
           └─► video_analyze ──► scdet scene detection ──► shot boundaries
                                 ├─► ffmpeg frame sampling (one representative frame per shot, capped)
                                 ├─► ffmpeg audio extract ──► ASR transcript (timestamped)   [optional]
                                 └─► OpenAI-compatible vision API ──► evidence JSON

• Scene changes are detected with ffmpeg's scdet filter (ffmpeg ≥ 6.0). Videos without detectable cuts fall back to uniform midpoint sampling.
• ASR is strictly additive: if asrApiKeyEnv is unset or the provider fails, the visual analysis still completes and transcript is null.
• All media work is delegated to ffmpeg/ffprobe on PATH — no native decoding in the agent.

Install

Prerequisites: Node.js ≥ 20, ffmpeg ≥ 6.0 (recommended) with ffprobe on PATH (brew install ffmpeg / apt install ffmpeg).

Option A — npm (recommended)

# in your DSH profile directory (the one containing package.json)
pnpm add dsh-video-lens

Option B — from source (development)

Clone the repo, then mount it into your DSH profile via a local link:

git clone https://github.com/dundunhan/dsh-video-lens.git

Either way, register the bundle in your profile's package.json — this exact block is the full profile configuration:

{
  "dependencies": {
    "dsh-video-lens": "^0.3"
  },
  "dsh": {
    "profile": {
      "bundles": [
        "@deepseek-ai/dsh-base",
        "@deepseek-ai/dsh-web-app",
        "dsh-video-lens"
      ]
    }
  }
}

Then export the keys and restart the profile:

export VIDEO_LENS_API_KEY=sk-...        # vision
export VIDEO_LENS_ASR_KEY=sk-...        # optional, ASR

Do not install the host runtime yourself. @deepseek-ai/dsh-tools is declared as an optional peer: the plugin always uses the dsh-tools that already ships with your DSH installation / DSH Desktop. Adding it to your profile as a dependency — or pinning one exact -rc version, which is what 0.3.1 did — installs a second, older runtime next to the host's, makes the Loader entry fail to import, and takes the whole plugin tree (and the app) down with it.

Configuration

All options are DSH config values:

Key	Default	Meaning
visionBaseUrl	https://api.siliconflow.cn/v1	Vision endpoint (OpenAI-compatible)
visionModel	Qwen/Qwen3-VL-8B-Instruct	Vision model name
visionApiKeyEnv	VIDEO_LENS_API_KEY	Env var holding the vision key
asrBaseUrl	https://api.siliconflow.cn/v1	ASR endpoint (OpenAI-compatible /audio/transcriptions)
asrModel	FunAudioLLM/SenseVoiceSmall	ASR model name
asrApiKeyEnv	VIDEO_LENS_ASR_KEY	Env var holding the ASR key
maxFrames	12	Frame budget cap (1–max); actual count is duration-adaptive (~1 frame per 30s, denser for short videos)
frameMaxWidth	768	Max frame width; keeps payloads small
frameQuality	4	JPEG quality (ffmpeg -q:v)
sceneThreshold	10	scdet threshold (0–100); higher = fewer cuts
askPaddingSec	2	video_ask window padding around matched transcript segments
vlmMaxTokens	1500	Vision model max output tokens
vlmTimeoutMs	90000	Vision call timeout
asrTimeoutMs	120000	ASR call timeout

Usage

Ask the agent:

"What's in /tmp/demo.mp4?"

The agent calls video_probe first, then video_analyze. Evidence includes:

{
  "metadata": { "container": "mov,mp4,m4a,3gp,3g2,mj2", "durationSec": 268.4, "...": "..." },
  "shots": [{ "timeSec": 12.3, "score": 45.2 }],
  "framesSampled": [{ "timestampSec": 5.5, "jpegBytes": 12345 }],
  "transcript": {
    "text": "…",
    "segments": [{ "start": 0.0, "end": 2.4, "text": "…" }],
    "language": "zh"
  },
  "visionModel": "Qwen/Qwen3-VL-8B-Instruct",
  "analysis": { "overall_summary": "…", "timeline": [{"timestamp_sec": 5.5, "description": "…"}], "on_screen_text": "…", "visual_style": "…", "notable_moments": "…" }
}

Permissions & security

Read this before using or redistributing. DSH plugins run in the host process as trusted code and there is no official plugin review — self-review is on the author. See SECURITY.md.

What this plugin does

• Reads: any local file path the agent passes to its tools (via ffprobe/ffmpeg).
• Executes: ffprobe and ffmpeg from PATH (never a shell — argv arrays only).
• Network: one outbound call per video_analyze to the configured visionBaseUrl (frames + vision key), and optionally one to asrBaseUrl (audio + ASR key).
• Does not: execute shells, eval code, phone home, auto-update, or read files on its own.

Operator responsibilities

• Keys are only as safe as the endpoints they are sent to — configure only endpoints you trust.
• The real access boundary is the DSH host sandbox; the plugin's readability check is a UX guard, not a security boundary.
• Payload sizes are bounded: maxFrames × ~100–300 KB (768px JPEG) per analysis call.

Compatibility

• Tested with DSH profile bundles @deepseek-ai/dsh-base + @deepseek-ai/dsh-web-app.
• Host runtime is not pinned: @deepseek-ai/dsh-tools is an optional peer resolved from the host installation, so the plugin follows the core it is loaded by (verified against core 0.1.0-rc.7 and 0.1.5-rc.2, the upstream version DSH Desktop 2.0.5 pins).
• Node ≥ 20 (uses AbortSignal.any / built-in fetch / FormData).
• ffmpeg ≥ 6.0 for scdet; older versions degrade to uniform sampling.
• macOS verified. Windows: the 0.3.1 boot failure reported on Windows was not platform-specific — it was the pinned old dsh-tools runtime (see Troubleshooting); the code paths themselves are OS-neutral (ffmpeg/ffprobe are spawned via argv, no shell).

Troubleshooting

dsh-plugin-desktop: plugin tree failed to load: failed to apply loader entry include (cordis:include): AggregateError: loader entries failed to apply — the client no longer starts.

0.3.1 hit this on DSH Desktop 2.0.5. The full error underneath is an import failure of the plugin (or of the host tools entry):

failed to import loader entry video-lens (dsh-video-lens): The requested module '@deepseek-ai/dsh-llm' does not provide an export named 'CallId'
  [cause]: profiles/<name>/node_modules/@deepseek-ai/dsh-tools/lib/index.js:4

Cause: 0.3.1 pinned @deepseek-ai/[email protected], so the profile got a second, older dsh-tools while the host ran a newer core (0.1.5-rc.2). Any failing Loader entry fails the whole tree, so the app cannot boot until the plugin is removed.

Recovery (0.3.1 installed and the app will not start):

1. Use the client's Recovery page to return to the last healthy profile, or remove the plugin from the profile: dsh plugin --profile <name> remove dsh-video-lens (Desktop: run that in its terminal).
2. Install dsh-video-lens@^0.3.2, where the host runtime is an optional peer and nothing is installed into the profile.

Uninstall

1. Remove dsh-video-lens from dsh.profile.bundles in your profile package.json.
2. Remove the dependency: pnpm remove dsh-video-lens (npm install) — or delete the link: entry if you installed from source — then reinstall the profile.

Roadmap

• v1.0: frame caching by file hash, evaluation table in README (5 video types × metrics), publish to npm (in progress).
• Beyond: native video-input models as an optional fast path when the configured VLM supports them.

License

MIT — see LICENSE.