Gemini Image Generator

Generate, edit, and compose images using Gemini models. Activate when user asks to generate images, draw, create logos/posters/icons/banners, edit/modify pho...

This skill likely does what it says — it sends image-generation requests to a configured Gemini endpoint — but there are three practical risks to consider before installing: (1) SKILL.md and the script require GEMINI_API_KEY and GEMINI_BASE_URL even though the registry lists none; plan to provide only a dedicated API key, not a general-purpose credential. (2) The script loads .env files from /workspace/.env and your home ~/.openclaw/.env and will import every KEY=VALUE into the environment; check those files first (or avoid placing other secrets there) because the skill will read them. (3) the README recommends running a remote curl | sh to install 'uv' — avoid running untrusted install scripts; prefer installing uv from a known package source or reviewing the installer first. If you decide to proceed, run the skill in an isolated environment (or container), supply a dedicated Gemini API key, and review the full script source for any additional network calls or logging that might leak sensitive data.

Type: OpenClaw Skill Name: wangyan-gemini-image-gen Version: 1.0.2 The skill provides image generation and editing capabilities using Gemini models via OpenAI-compatible or Google-native APIs. The Python script `scripts/generate_image.py` implements robust API interaction, model fallback logic, and image processing using standard libraries like `httpx` and `PIL`. While the documentation in `SKILL.md` suggests `curl|bash` installation methods for dependencies (uv and skillhub), these are presented as standard utility installation patterns and do not show evidence of malicious intent. The code logic is transparent, focuses on the stated purpose, and follows OpenClaw conventions for secret management and media handling.