Claude Code Video Vision is a plugin that lets Claude Code inspect videos by extracting image frames and producing timestamped audio transcriptions. It supports YouTube URLs and uses Gemini, local Whisper, or OpenAI processing backends, while the catalogue entries provide its commands, agent, and skill.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/jordanrendric/claude-video-visionWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/jordanrendric/claude-video-vision/watch-video)<a href="https://agentmods.dev/commands/jordanrendric/claude-video-vision/watch-video"><img src="https://agentmods.dev/badge/commands/jordanrendric/claude-video-vision/watch-video/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/jordanrendric/claude-video-vision/watch-video"><img src="https://agentmods.dev/badge/commands/jordanrendric/claude-video-vision/watch-video.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00017 | $0.00471 |
| Opus 5 | $0.00009 | $0.00235 |
| Sonnet 5 | $0.00003 | $0.00094 |
| Haiku 4.5 | $0.00002 | $0.00047 |
Grade A, and why
watch-video scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Watch Video
Parse the user's input to extract:
- Video path — the file path (required)
- YouTube URLs are also supported; pass the URL directly as
path
- YouTube URLs are also supported; pass the URL directly as
- Prompt — any question or instruction about the video (optional)
- Flags —
--fps <number>,--resolution <number>(optional)
Then follow this workflow in order — do NOT skip step 2:
-
Call
video_infoon the path/URL to verify it's a valid video and get duration. For YouTube URLs, the MCP server downloads the video withyt-dlpand uses YouTube subtitles/auto-captions before falling back to the configured audio backend. -
REQUIRED for videos > 30s: Call
video_analyzeBEFOREvideo_watch. This is NOT optional. Use filters:scene_changes: true, silence: true, transcription: trueat minimum. Add other filters based on the user's question (motion, blur, exposure, loudness, etc.). The analysis tells you WHERE to look — use it to plan smart frame extraction. -
Call
video_watch:- Short videos (< 2 min): Use
fps: "auto"withoutview_sample— full coverage to avoid missing brief moments. - Long videos (> 2 min): Use
segmentswith variable FPS based on analysis data. Useview_sampleto limit initial frames.
- Short videos (< 2 min): Use
-
If the user asks for more detail on a specific moment, use
video_detailto drill in with higher FPS/resolution on a 3-5 second window. Useview_sample: 3first, then request specific timestamps. -
If the user provided a prompt/question, answer it based on the video content.
-
If no prompt was provided, give a comprehensive summary of what happens in the video.
If video_watch fails with a setup error, call video_setup first, then retry.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 35 lines · 17 tokens per session scan A b29014b1ea43
watch-video is a command published in the GitHub repository jordanrendric/claude-video-vision (1,296 stars, last pushed 1mo ago), licensed MIT. It adds 17 tokens to every session and 471 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
media
Understand an audio / video / image file — Antigravity (agy/Gemini) transcribes and analyzes it, returning a timestamped digest while the full transcript goes to a file.
media
Ask a question about an AUDIO, VIDEO or IMAGE file (or a YouTube/remote URL) with Antigravity (agy / Gemini 3.x multimodal). Beyond transcription — "what decisions were made in this meeting?", "what happens at 2:30 in the video?", "what's the tone of this voice note?". Claude Code can't see/hear media; agy can.
transcribe
Transcribe and summarize an AUDIO or VIDEO file (or a YouTube/remote URL) with Antigravity (agy / Gemini 3.x multimodal) — something Claude can't do natively. Voice notes, meetings, calls, screencasts. Faithful transcript + summary; timestamps for video/URLs. Saves to docs/agy/transcripts/.
video
WATCH a video and return a structured VISUAL breakdown (not just a transcript) with Antigravity (agy / Gemini 3.x, natively multimodal in video). Scene-by-scene segments with timestamps, on-screen text/OCR (slides, charts, UI, captions), key visual events, and a summary. For screencasts, tutorials, presentations…
explain-video
Bridge workflow — analyze content with Gemini research tools, then synthesize an explainer video.
explain-status
Check status of explainer video projects.