Gemini Image Remix

Generate or remix images using Gemini models with text prompts and multiple input images, supporting various styles, resolutions, and advanced model options.

Security Scan

The skill's requirements, instructions, and code are coherent with an image-generation/remix tool that uses a Gemini API key and a 'uv' runner; nothing in the bundle appears to request unrelated credentials or perform unexpected exfiltration.

Purpose & Capability

Name/description match the code and runtime instructions. Requiring GEMINI_API_KEY and the 'uv' runner is appropriate for a script that calls the Gemini API via google-genai and is executed with 'uv run'.

Instruction Scope

SKILL.md and the script instruct running the local Python script and passing local image paths and a prompt. The script reads only the input image files you explicitly provide, the prompt, and the GEMINI_API_KEY; it sends those to the Gemini API. It does not attempt to read unrelated system files or other environment variables. Note: transmitting local images and prompts to the Gemini service is expected behavior but may have privacy implications.

Install Mechanism

Install spec uses a Homebrew formula 'uv' (traceable and low risk). The Python dependencies (google-genai, pillow) are declared in a header comment in the script but not installed by the SKILL.md install step — you'll need to install them in your Python environment (pip). This is not malicious but is a potential operational gap to be aware of.

Credentials

Only GEMINI_API_KEY is required and designated as the primary credential, which is proportional to the skill's purpose. No other secrets or unrelated environment/config paths are requested.

Persistence & Privilege

The skill is not force-included (always: false) and does not request elevated or persistent system privileges. It does not modify other skills or system-wide agent settings.

Assessment

This skill will read any local image files you pass to it and send those images plus your text prompt to the Gemini API using the GEMINI_API_KEY. Before installing or running it: 1) Confirm you trust the Gemini API endpoint/service for handling the images and prompts (private or sensitive images may be exposed). 2) Provide a scoped or replaceable API key (rotate or revoke if needed). 3) Install the Python dependencies (google-genai, pillow) from official PyPI packages in a controlled environment. 4) Verify the 'uv' brew formula source on your system if you do not already have 'uv'. If you need the script to avoid uploading certain files, do not pass those files as input. Overall the package is consistent with its stated purpose; the main risks are privacy of uploaded images and managing the API key.

A versatile tool for text-to-image generation and complex image-to-image remixing. By default, it uses Gemini 2.5 Flash Image for fast, high-quality results. It also supports flagship models like Gemini 3.0 Pro (Nano Banana Pro) for advanced artistic tasks.

Generate Image

Create stunning visuals from a text prompt.

uv run {baseDir}/scripts/remix.py --prompt "a cybernetic owl in a neon forest" --filename "owl.png"

Remix/Modify Image

Use one or more reference images to guide the generation. Perfect for style transfers, background changes, or character modifications.

uv run {baseDir}/scripts/remix.py --prompt "change the art style to a pencil sketch" --filename "sketch.png" -i "original.png"

Multi-image Composition

Combine elements from up to 14 different images into a single cohesive scene.

uv run {baseDir}/scripts/remix.py --prompt "place the character from image 1 into the environment of image 2" --filename "result.png" -i "character.png" -i "env.png"

Advanced Model Selection

Switch to advanced models like Nano Banana Pro for high-fidelity work.

uv run {baseDir}/scripts/remix.py --model "gemini-3-pro-image-preview" --prompt "highly detailed oil painting of a dragon" --filename "dragon.png"

Options

• --prompt, -p: Image description or specific edit instructions.
• --filename, -f: The output path for the generated PNG.
• --input-image, -i: Path to an input image (repeatable up to 14 times).
• --resolution, -r: 1K (default), 2K, or 4K.
• --aspect-ratio, -a: Output aspect ratio (e.g., 1:1, 16:9, 9:16, 4:3, 3:4).
• --model, -m: Model to use (defaults to gemini-2.5-flash-image). Supported: gemini-2.5-flash-image, gemini-3-pro-image-preview.
• --api-key, -k: Gemini API key (defaults to GEMINI_API_KEY env var).