Nanobanana Pro

**Overall**: Solid, practical image-generation/editing skill built around the Gemini Image API, with a useful automatic model fallback chain. Well-suited for OpenClaw workflows that need reliable, scriptable image creation and editing, provided the environment supports `uv` and a valid Gemini API key. **Strengths** - Handles text-to-image, single-image edits, and multi-image (up to 14) composition in one script. - Automatic fallback from `gemini-2.5-flash-image` to `gemini-2.0-flash-exp-image-generation` improves robustness against model errors or temporary outages. - Clear CLI usage (`uv run` or wrapper), multiple resolution presets (1K/2K/4K), and timestamped filenames make it practical in automation. - `MEDIA:` line output integrates cleanly with OpenClaw to auto-attach generated images. **Risks / Limitations** - Hard dependency on `uv run` (or the provided wrapper) may be problematic in environments where `uv` is not installed or allowed. - Fully dependent on external Gemini APIs: outages, latency spikes, model changes, or quota/limits will directly impact behavior. - Safety is primarily delegated to Gemini’s own content filters; there are no extra, skill-level policy checks or guardrails. - Requires secure handling of `GEMINI_API_KEY` (via env vars or OpenClaw config); misconfiguration can lead to failures or unintended usage. **Recommended Scenarios** - OpenClaw-based assistants or tools that frequently generate or edit images and benefit from automatic model fallback. - Content creation, prototyping, or design workflows needing quick visual outputs at 1K/2K/4K resolutions. - Automated pipelines where images must be programmatically generated and then attached to conversations or tickets without manual intervention. - Users comfortable with CLI and environment variable configuration who want a robust, low-friction Gemini image frontend.