Craft Image Prompts: skill pack for Hermes Agent

Portable Agent Skill that turns a visual brief into a precise prompt for image generation or editing

What Craft Image Prompts does

Craft Image Prompts is a portable Agent Skill that turns a rough visual brief into a model-ready prompt. It covers generation, editing, compositing, variation and data-visualization requests, and adapts its approach to GPT Image, Gemini, Midjourney, Flux, Stable Diffusion or SDXL, Adobe Firefly, or an unknown model. It handles reference images, visual continuity, exact on-image text and negative prompts.

By default the output is one clean English prompt, or a structured JSON object holding the brief, constraints and a final_prompt_string. It avoids generic filler such as 'cinematic', adds technical detail only when it improves the result, and keeps factual charts honest by refusing to invent values, labels or sources. When a missing detail would change the result, it asks at most three questions. It is a prompting skill only, so you run the prompt in the image model you choose. The repository holds SKILL.md, a model-adaptation reference and Codex metadata.

Key features

Prompts for generation, editing, compositing, variation and data visualization
Model-specific adaptation for GPT Image, Gemini, Midjourney, Flux, SDXL and Firefly
Support for reference images, visual continuity, on-image text and negative prompts
Plain English prompt output or structured JSON with final_prompt_string
Asks at most three clarifying questions
No fabricated values, labels or sources in charts

When to use it

Write a product-shot prompt for a specific aspect ratio and image model
Rewrite an image-edit request so it preserves angle and lighting
Produce several prompt variations that keep one character and palette
Build an infographic prompt from verified data

Who it is for: People who write prompts for image models and want an agent to apply model-specific rules for them.

How it fits with Hermes Agent

The repository is tagged hermes-agent and hermes-skill and is built as a portable Agent Skill. Its README documents installation for Codex and ChatGPT, not Hermes.

How to install Craft Image Prompts

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

git clone https://github.com/jducerf/craft-image-prompts.git
cp -R craft-image-prompts ~/.codex/skills/

Note: The repository has no license file, and the README documents installation for Codex and ChatGPT only.

FAQ

What is Craft Image Prompts?

Craft Image Prompts is a portable Agent Skill that converts a visual brief into a precise prompt for image generation or editing, adapted to the model you name.

Does Craft Image Prompts work with Hermes Agent?

The repository is tagged hermes-agent and hermes-skill and uses the portable SKILL.md format. The README only documents installing it in Codex and ChatGPT, so check how your Hermes setup loads skills.

Is Craft Image Prompts free and open source?

The repository has no license file, so default copyright applies and you should check with the author before reuse.