Image Generation
This is a skill for generating and reviewing relevant images using Generative AI MCPs
- Category: discovered
- Tags: github, community
- Version: 1.0.0
- Author: walid
- Downloads: 0
- Updated: 2025-11-22
Download the latest version (v1.0.0) from this page.
SKILL.md
--- name: Image Generation description: This is a skill for generating and reviewing relevant images using Generative AI MCPs --- # Image Generation Generate brand-aligned visual assets using AI models. This skill provides structured workflows for creating images that match your brand identity, either through guideline-based validation or reference-image-driven generation. ## Instructions Choose between **pre-defined workflows** for brand-aligned generation with systematic validation, or use the **generic iterative process** below for quick exploration without brand constraints. ### Available Workflows | Feature | [On-Brand Visual](workflows/on-brand-visual/WORKFLOW.md) | [Brand-Aware Asset Generation](workflows/brand-aware-asset-generation/WORKFLOW.md) | |---------|-----------------------------------------------------------|------------------------------------------------------------------------------------| | Brand context input | Visual guidelines from strategy | Visual assets as image references | | Upload requirement | No upload (text-only guidelines) | Yes (uploads brand assets to cloud) | | Validation approach | Systematic testing against guidelines | Implicit brand influence via image input | | Feedback structure | Dual tests (brand + user) per iteration | Single user feedback loop | | Draft tracking | Numbered suffix per iteration | Iteration counter | | Primary use case | Guideline-driven validation | Reference-image-driven generation | **When to use each:** - **On-Brand Visual:** Have written brand guidelines, need systematic validation - **Brand-Aware Asset Generation:** Have visual reference materials, need visual consistency **For quick exploration:** Continue with the basic process below. You are tasked with generating an image according to a user prompt. You will trigger an iterative self-optimizing 5 step image process. ### Process 1. Develop an optimized image `Prompt` to match the `Request` of the user. You will create a PROMPT.md file to input your prompt, leveraging the [`prompt-guide.md`](prompt-guide.md) . 2. Utilise the Replicate MCP and best model to generate an image and receive a public URL of the generated image. 3. Download the image to a local /tmp folder 4. View the image with your Read tool 5. Assess if the image adequately matches the `Request` Make a `Decision`: 1. if it matches, and meets expectations return a description of the image 2. If it fails to match, update your initial `Prompt` This will loop until the match condition is met. ## Available models ## Examples ### Example: Brand-Aware Generation User: "Create a hero image for our landing page" Agent approach: 1. Invoke brand-aware workflow: workflows/brand-aware-asset-generation/WORKFLOW.md 2. Workflow searches /brand/strategy/ for brand assets 3. Uploads found assets to cloud storage 4. Enhances prompt with brand context 5. Generates using google/nano-banana with brand references 6. Iterates based on user feedback (max 5 iterations)