
gemini-skills
Google Gemini 3 Pro skills for image generation (Nano Banana Pro) and AI consultation with web search and vision
- version
- 1.0.4
- GitHub stars
- 7
Install
/plugin install gemini-skills@gemini-claude-skillsRun it inside Claude Code after adding the marketplace with /plugin marketplace add stared/gemini-claude-skills.
Contents
Discovered from sourceSkills 2
Gemini Consultant
Get a second opinion from Google's Gemini 3 Pro (gemini-3-pro-preview) with real-time Google Search grounding.
CRITICAL: GEMINI IS BLIND. It knows nothing about your current session, file system, conversation history, or previous errors unless you explicitly force-feed it that data.
The "Dump Everything" Rule
The -c/--context argument is effectively unlimited. To get a useful answer, you must dump EVERYTHING relevant into the context.
- Quantity: We are talking about 10, 100, or 500+ lines of context.
- Scope: Include the file in question, all related files (imports, types, configs), the full error traceback, your current hypothesis, and a summary of what you have already tried.
- History: If you are deep in a debugging session, dump the relevant parts of the conversation history into the context.
If you think a file is even tangentially related, INCLUDE IT.
Prerequisites
GEMINI_API_KEY environment variable must be set.
Usage
uv run /path/to/skills/gemini-consultant/consult.py "Detailed Question" -c "MASSIVE CONTEXT STRING"
Parameters
| Parameter | Description |
|---|---|
question | The prompt. Be specific. "Given error X and code Y (in context), why is logic Z failing?" |
-c, --context | THE DATA DUMP. Concatenate everything here. Code, logs, history, configs. Can be used multiple times. |
-i, --image | Image file paths (screenshots, diagrams). |
--media-resolution | low, medium (default), high, ultra_high. |
--no-search | Disable web search (pure code logic). |
--thinking | low or high (default). Use high for code. |
Examples
BAD (Lazy = Failure)
# WRONG: Context is missing.
uv run consult.py "Why is this crashing?" -c "Error: 500 Internal Server Error"
# WRONG: Single file provided, ignoring dependencies.
uv run consult.py "Fix this function." -c "$(cat src/broken_file.py)"
GOOD (Exhaustive = Success)
NOTE: Real-world usages are usually 10-100x longer than the snippets below.
# CORRECT: Providing File, Imports, Types, and Errors
uv run consult.py "I am getting a RecursionError in the serialization logic. Analyze the relationship between the Model and the Schema." \
-c "FILE: src/models.py
$(cat src/models.py)
FILE: src/schemas.py
$(cat src/schemas.py)
FILE: src/config/db_settings.py
$(cat src/config/db_settings.py)
ERROR LOG:
$(cat logs/error_dump.txt)
PREVIOUS ATTEMPTS:
I tried removing the circular reference in schema.py line 40 but it broke the API validation."
# CORRECT: Architecture review with full project scope
uv run consult.py "Review this authentication flow for security flaws." \
-c "MIDDLEWARE: $(cat src/middleware/auth.ts)
MODELS: $(cat src/models/user.ts)
ROUTES: $(cat src/routes/auth.ts)
CONFIG: $(cat src/config/security.ts)
TYPES: $(cat src/types/express.d.ts)"
Web Search (No Code Context Needed)
uv run consult.py "What is the latest version of Next.js and what are its new features?"
Nano Banana Pro Image Generator
Generate images using Google's advanced Nano Banana Pro model (gemini-3-pro-image-preview).
Prerequisites
The user must have GEMINI_API_KEY environment variable set with a valid Google AI API key.
Usage
The script is located in the same directory as this SKILL.md file. Run it with uv run:
uv run /path/to/skills/nano-banana-pro/generate_image.py "your prompt" -o output.png
When this skill is invoked, locate generate_image.py in the skill directory and run it.
Parameters
| Parameter | Required | Description |
|---|---|---|
prompt | Yes | Text description of the image to generate or transformation to apply |
-o, --output | Yes | Output filename (you decide the path based on context) |
-i, --image | No | Input image(s) for editing/transformation or as context/reference (can be used multiple times) |
--aspect-ratio | No | One of: 1:1, 2:3, 3:2, 3:4, 4:3, 4:5, 5:4, 9:16, 16:9, 21:9 (default: 1:1) |
--size | No | Image size: 1K, 2K, 4K (default: 1K) |
Examples
Basic image generation:
uv run generate_image.py "A sunset over mountains" -o sunset.png
Infographic with specific aspect ratio:
uv run generate_image.py "Infographic showing the water cycle with labeled stages" -o water_cycle.png --aspect-ratio 9:16
High-resolution ultrawide:
uv run generate_image.py "Professional photo of a modern office space" -o office.png --aspect-ratio 21:9 --size 4K
Edit an existing image:
uv run generate_image.py "Add a sunset sky to this image" -i photo.png -o edited.png
Transform with style:
uv run generate_image.py "Make this look like a watercolor painting" -i input.jpg -o watercolor.png
Combine multiple images:
uv run generate_image.py "Create a collage blending these images together" -i img1.png -i img2.png -o collage.png
Use image as context/reference:
uv run generate_image.py "Generate a new landscape in the same style as this reference" -i reference.png -o new_landscape.png
Model Capabilities
Nano Banana Pro excels at:
- Accurate infographics with real data (uses Google Search grounding)
- Text rendering in images
- Image editing and transformation from input images
- Context-aware generation using reference images for style, composition, or subject
- Cartographic visualizations and maps
- Detailed instruction following
- Chain-of-thought reasoning for complex visual tasks
Output
The script prints:
- Progress message while generating
- Path to saved image on success
- Any text response from the model
- Error message if no image was generated
Sponsors
Details
- Marketplace
- stared/gemini-claude-skills
- Category
- Community