Gemini Skills

To use the Gemini Interactions API, install the appropriate SDK for your language. Note that legacy SDKs (google-generativeai for Python and @google/generative-ai for JS) are deprecated. Use the following versions:

• Python: Install google-genai version 2.3.0 or higher.
• JavaScript/TypeScript: Install @google/genai version 2.3.0 or higher.

# Python
pip install -U google-genai

# JavaScript/TypeScript
npm install @google/genai

Gemini Omni Flash supports high-quality text rendering. You can define text that appears on screen or naturally occurring text in the environment.

Examples:

• One word on the screen at a time: "did, you, know, that, Omni, can, do, awesome, text?" Each word appears for 1s with a different animated style. No dialogue.
• There is a street sign that says: "This is an AI generation by Omni", there is a storefront that says: "All you need AI", there's a car with the number plate: "OMN111"

To guide Gemini Omni Flash on how to use uploaded media, use specific tags in your prompt to define whether an image is a starting frame or a reference.

Simple Tags

• <FIRST_FRAME>: Sets the image as the starting frame of the video.
• <IMAGE_REF_N>: Sets the image as a reference (where N is the index starting from 0).

Explicit Declaration

For complex prompts, use prefix tags to map images to roles:

• [# Sources <FIRST_FRAME>@Image1]
• [# References <IMAGE_REF_0>@Image1]

Example: [# Sources <FIRST_FRAME>@Image1] [# References <IMAGE_REF_0>@Image2] a woman <IMAGE_REF_0> is walking. Use Image1 as the starting frame. Use Image2 as a reference for the video generation.

The model generates an appropriate audio track by default. To control the audio or include specific music, describe the desired sound in your prompt using formats like:

• Include calm background music
• The video has a high energy techno beat
• The audio is a low tinny radio broadcast in the background, playing a song
• Audio design: [a description of the audio you want]

The Antigravity agent (antigravity-preview-05-2026) is a general-purpose managed agent that can execute code (Bash, Python, Node.js), manage files, and browse the web. To use it, you must set environment="remote" to provision a sandboxed Linux environment.

Python Example:

from google import genai

client = genai.Client()

interaction = client.interactions.create(
    agent="antigravity-preview-05-2026",
    input="Write a Python script that generates the first 20 Fibonacci numbers and saves them to fibonacci.txt. Then read the file and print its contents.",
    environment="remote",
)

print(f"Environment ID: {interaction.environment_id}")
print(interaction.output_text)

Use upload_file.py to upload local images or videos to the Files API. The script polls the API until the file status is ACTIVE. For videos larger than 25MB, it is recommended to use prep_video.py first to optimize the file for Gemini Omni Flash (which is optimized for 10s videos at 720p/24fps).

./scripts/upload_file.py path/to/image.png

You can use the Vercel skills CLI (skills.sh) to browse or install specific skills from this repository.

To interactively browse and list available skills:

npx skills add google-gemini/gemini-skills --list

To install a specific skill (for example, gemini-interactions-api):

npx skills add google-gemini/gemini-skills --skill gemini-interactions-api

To use the Gemini Live API, install the modern SDKs. Note that legacy SDKs (google-generativeai for Python and @google/generative-ai for JS) are deprecated.

Python: Install google-genai via pip.

JavaScript/TypeScript: Install @google/genai via npm.

# Python
pip install google-genai

# JavaScript/TypeScript
npm install @google/genai

To avoid specific elements in your video, you can include simple negative prompts in your request, such as:

• No dialogue
• No embellishments
• No extra sound effects