Guide · Agent skill
How to use the Image Generation agent skill (hive.image-generation)
Updated · written with AI from the project's own documentation and checked against our catalogue data
The Image Generation agent skill is a set of instructions that teaches an AI agent how to create and edit images from a written description using the image_generate tool. It is written for people who may never have used agent skills before, and this guide explains what the skill does, how to install it, how to ask for images in plain language, and what to watch for. By the end you will know how to get a first image, how to edit a reference photo, and how to handle the common failure messages.
Image Generationaden-hive/hive · Apache-2.0 Download the skill Quick start
- 1Download the skill. Download the skill ZIP from this page and unzip it so you have a folder containing SKILL.md.
- 2Place it in a skills folder. Move the folder to ~/.claude/skills/ for personal use or to .claude/skills/ inside your project.
- 3Or upload it in the Claude apps. In the Claude apps, open Settings → Capabilities → Skills and upload the skill.
- 4Make a request. Describe the image you want, including subject, style, colours and any text, in plain language.
- 5Add reference images if editing. Attach a photo or give its path so the agent can restyle or combine it.
- 6Wait for the result. The agent checks the background job until the image is ready, then attaches it in chat.
- 7Refine. Ask for changes such as a different colour or a bolder look, and the agent will generate a new version.
What the skill makes the agent do
The skill teaches the agent to call image_generate, which turns a text prompt into an image and can also edit or restyle existing images you supply as reference files or URLs. It routes requests through the Hive image service to OpenAI's gpt-image-2 model.
Costs are billed to your Hive credits in the same way as an AI text call, so no separate API key is needed. Each generated image is also saved to disk.
- Create logos, illustrations, icons, banners, posters, thumbnails, mockups, product shots and social graphics from a prompt
- Edit, restyle or combine existing images using reference images
- Show the finished image in chat as a downloadable file
Install the skill
Download the skill ZIP from this page and unzip it. The skill is a single file named SKILL.md. Put the unzipped folder in one of two places: your personal skills folder at ~/.claude/skills/, or a project-specific .claude/skills/ folder inside your project.
If you use the Claude apps instead, open Settings, go to Capabilities, then Skills, and upload the skill there.
- Personal: ~/.claude/skills/
- Project: .claude/skills/ inside your project folder
- Claude apps: Settings → Capabilities → Skills
Ask for an image in plain language
You do not need to name the skill or the tool. Requests such as making a picture, a logo, an illustration, an icon, a banner, a poster, a thumbnail, a hero image, a mockup, a product shot or a social graphic should trigger it. Asking to edit, restyle or combine an image you have shared should also trigger it.
The more concrete your request, the better. Name the subject, the style (photo, flat vector, 3D, watercolor), the framing, the colours, the mood, and any text that should appear in the image, quoted exactly.
- Good: "A minimalist bee logo, flat vector, amber on white"
- Good: "Put this product on a marble kitchen counter, soft morning light" (with the photo attached)
- Weak: "make something cool"
Understand the settings and cost
The defaults are set for low cost and speed. Image quality is locked to low, and medium and high are disabled for cost control. Any request for a higher tier is forced to low. The default size is 1024x1024, and landscape, portrait and auto sizes are also available.
The agent only produces several images when you explicitly ask for variations, because each image is billed separately.
- quality: low only (default)
- size: 1024x1024, 1536x1024 (landscape), 1024x1536 (portrait) or auto
- n: 1 to 4 images, each billed separately
- output_format: png, jpeg or webp
Edit or combine existing images
To edit a photo, attach it to your message or give its path. The agent passes it as a reference image. You can provide up to 10 local file paths or http(s) URLs. Images the agent generated earlier can also be reused, using their saved path.
- Restyle a product photo
- Place a logo on a mockup
- Keep a character's look across several images
- Merge elements from two or more images
image_generate(prompt="Put this product on a marble kitchen counter, soft morning light",
reference_images=["data/uploads/bottle.png"])What happens while an image is being made
Generation runs in the background because it can take a couple of minutes. The agent starts the job, receives a handle, and then checks on it with collect_result until the finished image is returned. Each check waits up to 45 seconds. A pending status just means the image is not ready yet.
When the image is ready, the agent attaches it to the chat so you can download it. It does not paste raw image data into the reply.
- Start: image_generate returns a handle such as bg_1
- Poll: collect_result returns pending until the image is done
- Finish: the image is attached in chat
start = image_generate(prompt="A minimalist bee logo, flat vector, amber on white")
res = collect_result(handle="bg_1", wait_seconds=30)Know the failure messages
Errors are reported in the result rather than stopping the agent. The skill tells the agent how to respond to each one. Out of credits or an inactive subscription means you need to add Hive credits, and the agent will not retry. Model unavailability is reported and not retried. A moderated or rejected prompt is usually rephrased once, with less explicit wording and no real-person likeness. Rate limits lead to one retry after a short wait.
If a job stays pending for several minutes, the agent will tell you it likely failed and start once more.
- 402: out of Hive credits or subscription inactive
- 403: image generation currently unavailable
- 400: prompt refused or moderated; rephrase
- 429: rate limited; retry once after a wait
- Pending for several minutes: likely failed; start again
What people use it for
- Quick logo concepts for a brand or side project
- Banner, thumbnail or social graphic drafts for a post
- Rough mockups that place a logo or product on a reference photo
- Simple illustrations or icons without opening an image editor
- Restyling an existing product photo onto a new background
Frequently asked questions
How do I install the image generation agent skill?
Download the skill ZIP from this page and unzip it. Place the folder in ~/.claude/skills/ for personal use or in .claude/skills/ inside a project. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills.
Do I need an API key to use it?
No. The skill runs through the Hive image service, and costs are billed to your Hive credits like an AI text call. There is no API key to configure.
Is the image generation skill free?
The source does not list pricing. Generation is billed to your Hive credits, and the skill keeps quality at the cheapest setting by default. Check your Hive credit balance for what you will be charged.
Why is the image quality always low?
The skill locks quality to low to control cost, and medium and high are disabled. Any request for a higher tier is forced to low, so it is not suited to final high-resolution print work.
Can it edit my own photos?
Yes. Pass existing images as reference images, either as local file paths or http(s) URLs, up to 10 per request. It can restyle, place a logo on a mockup, or merge elements.
What happens when I run out of Hive credits?
The skill tells the agent to stop and report that you are out of credits, and not to retry. Add credits to your Hive account before generating more images.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits marketers who need quick logo concepts, banners or social graphics, and designers who want rough mockups or product shots on reference photos. It is not suited to final high-resolution print work, to generating images of real people's likeness, or to situations where Hive credits are unavailable. Choose a dedicated image editor if you need precise control or high-quality output.
Download the skill