Skip to content

Latest commit

 

History

7 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

image-gen

A provider-neutral skill for generating images and delivering them as verified files.

It does not own model access. Whatever image-capable model your harness can already dispatch is the model it uses. The skill adds the parts that are easy to get wrong: prompt craft, candidate discipline, artifact verification, provenance and release QA.

Requirements

  • Python 3.10+
  • Pillow
python3 -m pip install -r requirements.txt

No endpoint or credential configuration. scripts/*.py never make a network request and never read a credential.

Use

  1. Ask your harness's image-capable model to produce the image.
  2. Save the payload as a verified artifact:
python3 scripts/save_image.py \
  --input response.json \
  --output out/hero.png \
  --prompt "a blue circle on white" \
  --model "<model id>"

--input accepts an images API response (data[0].b64_json), a payload with b64_json / image_base64 / b64, or a raw image file. The tool writes the image plus <image>.json provenance and refuses to overwrite unless --overwrite is passed.

Tools

Tool Purpose
scripts/save_image.py Verify a payload; write image + provenance
scripts/inspect_image.py Format, dimensions, mode, size, SHA-256
scripts/make_variants.py Deterministic crops; keeps alpha; refuses overwrite

Verify

python3 -m unittest discover -s tests -v

Tests are local and require no credential and no network.

About

Complex AIFlow image production, iteration and release QA skill

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages