Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generate or edit images using AI models (FLUX, Gemini). Use for scientific illustrations, diagrams, schematics, infographics, concept visualizations, and artistic images. Supports image editing to modify existing images (change colors, add/remove elements, style transfer). Useful for figures, posters, and visual explanations.
.claude/skills/microck-generate-image/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 80% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 86% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -37% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 18% | 0% |
Generate and edit high-quality images using OpenRouter's image generation models including FLUX.2 Pro and Nano Banana Pro (Gemini 3 Pro).
Use the scripts/generate_image.py script to generate or edit images:
bash# Generate a new image python scripts/generate_image.py "A beautiful sunset over mountains" # Edit an existing image python scripts/generate_image.py "Make the sky purple" --input photo.jpg
This generates/edits an image and saves it as generated_image.png in the current directory.
CRITICAL: The script requires an OpenRouter API key. Before running, check if the user has configured their API key:
.env file in the project directory or parent directoriesOPENROUTER_API_KEY=<key> in the .env file.env file with OPENROUTER_API_KEY=your-api-key-hereexport OPENROUTER_API_KEY=your-api-key-hereThe script will automatically detect the .env file and provide clear error messages if the API key is missing.
Default model: google/gemini-3-pro-image-preview (high quality, recommended)
Available models for generation and editing:
google/gemini-3-pro-image-preview - High quality, supports generation + editingblack-forest-labs/flux.2-pro - Fast, high quality, supports generation + editingGeneration only:
black-forest-labs/flux.2-dev - Development version, generation onlySelect based on:
bashpython scripts/generate_image.py "Your prompt here"
bashpython scripts/generate_image.py "A cat in space" --model "black-forest-labs/flux.2-pro"
bashpython scripts/generate_image.py "Abstract art" --output artwork.png
bashpython scripts/generate_image.py "Make the background blue" --input photo.jpg
bashpython scripts/generate_image.py "Add sunglasses to the person" --input portrait.png --model "black-forest-labs/flux.2-pro"
bashpython scripts/generate_image.py "Remove the text from the image" --input screenshot.png --output cleaned.png
Run the script multiple times with different prompts or output paths:
bashpython scripts/generate_image.py "Image 1 description" --output image1.png python scripts/generate_image.py "Image 2 description" --output image2.png
prompt (required): Text description of the image to generate, or editing instructions--input or -i: Input image path for editing (enables edit mode)--model or -m: OpenRouter model ID (default: google/gemini-3-pro-image-preview)--output or -o: Output file path (default: generated_image.png)--api-key: OpenRouter API key (overrides .env file)The script provides clear error messages for:
If the script fails, read the error message and address the issue before retrying.
images and content response formats from different OpenRouter models| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,524 | 4,917 | -11% | 1 | 1 | 0% | 956 | 1,336 | +40% | 0 | 0 | — |
case-02 | fail→fail | 7,116 | 6,429 | -10% | 1 | 1 | 0% | 1,222 | 1,471 | +20% | 0 | 0 | — |
case-03 | fail→pass | 11,090 | 2,955 | -73% | 1 | 1 | 0% | 1,658 | 1,564 | -6% | 0 | 0 | — |
case-04 | fail→pass | 5,187 | 2,538 | -51% | 1 | 1 | 0% | 876 | 1,580 | +80% | 0 | 0 | — |
case-05 | fail→pass | 4,918 | 1,519 | -69% | 1 | 1 | 0% | 726 | 1,349 | +86% | 0 | 0 | — |
case-06 | fail→pass | 12,649 | 1,744 | -86% | 1 | 1 | 0% | 2,281 | 1,435 | -37% | 0 | 0 | — |
case-07 | fail→pass | 6,092 | 1,555 | -74% | 1 | 1 | 0% | 1,106 | 1,302 | +18% | 0 | 0 | — |
case-08 | pass→pass | 6,348 | 2,287 | -64% | 1 | 1 | 0% | 997 | 1,420 | +42% | 0 | 0 | — |
case-09 | pass→pass | 7,022 | 2,014 | -71% | 1 | 1 | 0% | 1,183 | 1,420 | +20% | 0 | 0 | — |
case-10 | fail→pass | 9,437 | 1,296 | -86% | 1 | 1 | 0% | 1,553 | 1,278 | -18% | 0 | 0 | — |
case-19 | pass→pass | 10,207 | 1,432 | -86% | 1 | 1 | 0% | 1,569 | 1,298 | -17% | 0 | 0 | — |
case-11 | pass→pass | 13,251 | 3,699 | -72% | 1 | 1 | 0% | 2,384 | 1,722 | -28% | 0 | 0 | — |
case-12 | pass→pass | 4,568 | 1,803 | -61% | 1 | 1 | 0% | 693 | 1,345 | +94% | 0 | 0 | — |
case-13 | pass→pass | 13,333 | 3,328 | -75% | 1 | 1 | 0% | 2,250 | 1,685 | -25% | 0 | 0 | — |
case-14 | fail→pass | 11,155 | 2,119 | -81% | 1 | 1 | 0% | 1,184 | 1,494 | +26% | 0 | 0 | — |
case-15 | fail→pass | 8,935 | 4,337 | -51% | 1 | 1 | 0% | 1,555 | 1,897 | +22% | 0 | 0 | — |
case-16 | fail→pass | 4,696 | 2,143 | -54% | 1 | 1 | 0% | 762 | 1,451 | +90% | 0 | 0 | — |
case-17 | pass→pass | 15,210 | 9,101 | -40% | 1 | 1 | 0% | 2,341 | 2,740 | +17% | 0 | 0 | — |
case-18 | pass→pass | 10,543 | 2,307 | -78% | 1 | 1 | 0% | 1,707 | 1,417 | -17% | 0 | 0 | — |
case-20 | fail→pass | 9,587 | 1,931 | -80% | 1 | 1 | 0% | 1,657 | 1,461 | -12% | 0 | 0 | — |
case-21 | pass→pass | 6,425 | 2,244 | -65% | 1 | 1 | 0% | 1,005 | 1,487 | +48% | 0 | 0 | — |
case-22 | pass→pass | 3,919 | 1,662 | -58% | 1 | 1 | 0% | 635 | 1,303 | +105% | 0 | 0 | — |
case-23 | pass→pass | 9,070 | 6,031 | -34% | 1 | 1 | 0% | 1,757 | 2,341 | +33% | 0 | 0 | — |
case-24 | pass→pass | 7,843 | 4,924 | -37% | 1 | 1 | 0% | 1,607 | 2,064 | +28% | 0 | 0 | — |
case-25 | pass→pass | 10,160 | 6,189 | -39% | 1 | 1 | 0% | 1,750 | 2,335 | +33% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 25 cases were attempted, and 23 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +40 percentage points is the difference between those two pass rates over the 23 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.