Add asset-generation skill for image generation via Gemini

2026-03-03 17:04:48 +01:00
parent 9d8e0e1024
commit 10303dad09
2 changed files with 99 additions and 0 deletions
@@ -0,0 +1,38 @@
+# asset-generation
+
+Generate images from text prompts using the Gemini API.
+
+## Setup
+
+### 1. Get a Gemini API key
+
+1. Go to [Google AI Studio](https://aistudio.google.com/apikey)
+2. Create a new API key
+
+### 2. Configure the key
+
+Add to your `.env.tools` file in the project root:
+
+```
+GEMINI_API_KEY=your_key_here
+```
+
+Make sure `.env.tools` is in your `.gitignore` — never commit API keys.
+
+### 3. Verify it works
+
+```bash
+node asset-generation/scripts/generate-image.mjs \
+  --prompt "a simple blue circle on white background" \
+  --output assets/test.png
+```
+
+## Usage
+
+```bash
+node asset-generation/scripts/generate-image.mjs \
+  --prompt "description of the image" \
+  --output assets/my-image.png
+```
+
+See `SKILL.md` for the full workflow.
@@ -0,0 +1,61 @@
+---
+name: asset-generation
+description: >
+  Generate images from text prompts using Gemini. Triggers - generate an image, create an asset,
+  make a graphic, generate a logo, create an illustration, I need an image for.
+---
+
+# Asset Generation
+
+Generate images from text prompts using the Gemini model and save them to the project's `/assets` directory.
+
+## Prerequisites
+
+- Set `GEMINI_API_KEY` in a `.env.tools` file at the project root (see `asset-generation/README.md`).
+- Node.js installed.
+
+## When to use this skill
+
+Use when:
+
+- The user asks to **generate an image** (logo, illustration, graphic, placeholder, etc.)
+- The user says "create an asset" / "make a graphic" / "I need an image for..."
+- A visual asset is needed for the project and no existing file fits
+
+## Workflow
+
+### 1. Clarify the Prompt
+
+Before generating, make sure you have a clear prompt:
+- What is the image for? (icon, hero image, placeholder, illustration, etc.)
+- What style? (flat, realistic, minimal, etc.)
+- Any specific colors, dimensions, or constraints?
+- What filename should it have?
+
+If the user gave a clear enough description, proceed directly.
+
+### 2. Generate the Image
+
+```bash
+node asset-generation/scripts/generate-image.mjs \
+  --prompt "your detailed prompt here" \
+  --output assets/filename.png
+```
+
+The script will:
+- Send the prompt to the Gemini API
+- Save the generated image to the specified path
+- Create the `/assets` directory if it doesn't exist
+
+### 3. Verify the Output
+
+- Confirm the file was created at the expected path
+- Report the file path and size to the user
+- If the result doesn't match expectations, refine the prompt and regenerate
+
+## Rules
+
+- **Always confirm the prompt** with the user before generating (unless it's already specific)
+- **Use descriptive filenames** — `hero-banner.png` not `image1.png`
+- **Save to `/assets`** directory by default, unless the user specifies another location
+- **Don't overwrite** existing files without asking