Gemini Video Analyst

    1

    **Gemini Video Analyst** lets your agent watch videos for you.

    Free

    1 use5.0 (1 review)

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Gemini Video Analyst

    Example session with this skill installed

    What is said and what is shown in this video? Keep the two apart and add timestamps: https://www.youtube.com/watch?v=jNQXAC9IVRw

    • Read your context and instructions
    • Compiled the gemini video analyst

    Video: https://www.youtube.com/watch?v=jNQXAC9IVRw ·

    Mode: agentic ·

    Tokens: 3,963 ·

    Model: gemini-3.8-flash

    1. Summary

    A young man (Jawed Karim) stands outdoors in front of an elephant enclosure at a zoo and speaks directly to the camera about the elephants and their long trunks. He delivers a short, casual commentary about them being cool before concluding that there is little else to say.

    2. Timeline

    • 00:00 – 00:03: The speaker addresses the camera, introducing his location at the elephant exhibit.
    • 00:03 – 00:05: He turns his head to glance back over his shoulder at the elephants behind him.

    00:05 – 00:10: He turns back to face the camera, gestures with his right hand, and talks about the elephants' long trunks.

    • 00:11 – 00:14: He adds that "that's cool" and remarks that there is nothing much left to say.
    • 00:14 – 00:16: He turns his head once more to look back at the elephants.
    • 00:17 – 00:18: He turns back toward the camera, smiles slightly, and the video concludes.

    3. Spoken Content (What is Said)

    • 00:01 – 00:04: "All right, so here we are in front of the, uh, elephants."
    • 00:04 – 00:10: "And the cool thing about these guys is that, is that they have really, really, really..."
    • 00:10 – 00:12: "...long, um, trunks, and that's, that's cool."
    • 00:12 – 00:15: "And that's pretty much all there is to say."

    4. Visual Content (What is Shown but Not Said)

    Speaker: A young man with short dark hair, wearing a dark blue t-shirt under an open windbreaker jacket (red front panels, grey/black shoulders and trim).
    *

    Setting & Background: An outdoor zoo enclosure. To the left is a large artificial rock/cliff face. Behind the speaker is a barrier made of thick metal posts and horizontal steel cables.
    *

    Animals: Two elephants stand in the fenced pen in the background. The elephant directly behind the speaker has dry hay resting on its back and head, feeding from the ground/hay piles. A second elephant stands further to the right.
    *

    Camera / Visual Quality: Handheld camera held roughly at chest/eye level, showing slight natural hand movement; standard-definition/low-resolution digital video quality typical of early digital cameras/webcams (circa 2005).

    • On-screen Text: None.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Gemini Video Analyst lets your agent watch videos for you. Give it a public YouTube link, an Instagram reel or a video file on your computer and ask anything: a summary with chapters, every tool and link shown on screen, what a clip really claims, or the code visible in a terminal at 12:30.

    How it works

    The skill teaches your agent to use the open-source gemini-video MCP server (MIT licence, https://github.com/DerYUYU/gemini-video-mcp). The server sends the video to the Google Gemini API, which evaluates sound and picture together, so the answer covers what is shown and not only what is said.

    • Picks the cheapest mode: Gemini's agentic video mode for whole videos (measured on a 15-minute video: 7,416 instead of 85,480 tokens, with better coverage), exact time ranges and high resolution only when needed
    • Writes precise prompts in your language, with timestamps, keeping what is SAID apart from what is SHOWN
    • Handles long videos: anything over about 15 minutes is analyzed in sections of up to 5 minutes, and long local files are uploaded once, reused for every question, then cleaned up
    • Double-checks names when a fallback model answered, so misheard tool names don't reach you as facts
    • Walks you through the setup if the server isn't installed yet

    What you need

    Node.js 22 or newer, your own Gemini API key from Google AI Studio (the free tier allows about 20 requests per day per model, and the server falls back through several models automatically), and yt-dlp for Instagram only.

    Where it stops

    Public videos only: no private, unlisted or login-only content, no Instagram stories or image carousels, no TikTok. Local files: mp4, mov, webm, avi, wmv, mpeg, mpg, flv and 3gp up to 2 GB. Timestamps can be off by a few seconds. Not affiliated with Google.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Video Skills.

    Public YouTube videos, Instagram reels and video posts, and local video files up to 2 GB only. Private, unlisted, deleted and region-blocked videos fail, as do Instagram stories, private accounts and image carousels. TikTok and other platforms are not supported. Instagram depends on yt-dlp, which can break after Instagram changes its site until yt-dlp is updated. Timestamps can be off by a few seconds. On the free Gemini tier, Google may use submitted content to improve its products.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    5.0
    1 review
    5
    1
    4
    0
    3
    0
    2
    0
    1
    0

    This skill has been used 1 time.

    Trust & safety

    Security scan

    Security scan passed 22 days ago

    • Free with an Agensi account

    Software
    Gemini Video MCPNode.jsGityt-dlp

    Permissions

    Write Files
    Network Access
    Terminal / Shell
    Read Files

    The skill ships no scripts. Terminal is only used to install the gemini-video MCP server (git clone, npm install), and only after the user agrees. Network access happens inside the MCP server, which sends the video to the Google Gemini API with the user's own key. Local video files are read only when the user passes their path. The API key stays in the server's .env file; the skill never asks for it in the chat.

    Listed22 days ago

    What's inside

    Frequently Asked Questions

    Gemini evaluates the audio track and the picture together, so the answer also covers what only appears on screen: code, slides, product names, URLs and numbers.

    The free gemini-video MCP server (Node.js 22+), your own Gemini API key from Google AI Studio, and yt-dlp for Instagram only. The skill walks you through the setup if the server is not installed yet.

    Yes. Pass the absolute path of a local file (mp4, mov, webm, avi, wmv, mpeg, mpg, flv or 3gp, up to 2 GB). It is uploaded to the Gemini Files API once, reused for follow-up questions and kept for up to 48 hours unless you delete it earlier. On the free Gemini tier Google may use uploaded content to improve its products, so the skill asks before uploading anything private.

    Private, unlisted, deleted and region-blocked videos, Instagram stories, private accounts and image carousels, and other platforms such as TikTok.

    A SKILL.md with the workflow, mode choices, prompt recipes and error handling, plus a setup guide. The skill and the MCP server (MIT, on GitHub) are free. The Gemini API free tier is enough for everyday use; if you switch your key to a paid tier, Google bills you directly.

    No. The key goes into the server's .env file or your MCP client configuration. The skill tells the agent never to ask for it in the chat.

    New in Image & Video

    Free