ChatGPT Image Gen Skill

What ChatGPT Image Gen Skill does

The skill uses Playwright to connect to a Chromium-based browser (Chrome, Edge, Brave or Chromium) that is already logged in to ChatGPT and was started with --remote-debugging-port=9222. It sends your prompt to the ChatGPT web interface and then downloads the original image rather than a screenshot, saving it by default into ~/.hermes/chatgpt_images with a timestamped file name. Chinese and English prompts are supported, and a cinematic prompt template with examples is included.

You add the repository as a Hermes skill tap, search it with hermes skills search image and install chatgpt-image-gen. Hermes scans skills/*/SKILL.md in the repository. The script chatgpt_playwright.py can also run directly and accepts --cdp-url, --out-dir, --timeout and --prefix, with HERMES_CDP_URL and HERMES_CHATGPT_IMAGE_DIR as environment variables. The README gives start commands for the debugging port on macOS and Windows for Chrome and Edge. The documentation is bilingual in English and Chinese, and the license is MIT.

Key features

• Generates images through the ChatGPT web interface
• Downloads original image files instead of screenshots
• Chinese and English prompts
• Works with Chrome, Edge, Brave and other Chromium-based browsers
• Configurable CDP URL, output directory, timeout and prompt prefix

When to use it

• Asking Hermes for a cinematic concept image in English or Chinese
• Saving full-quality images locally from a ChatGPT session
• Running image generation from the command line with a single script

Who it is for: Hermes Agent users with a ChatGPT login who want Hermes to generate images through that account.

How it fits with Hermes Agent

Built for Hermes Agent as a skill that is distributed through a Hermes skill tap.

How to install ChatGPT Image Gen Skill

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

hermes skills tap add k1m0206/chatgpt-image-gen-skill
hermes skills install chatgpt-image-gen

Requirements: Python, Playwright (pip install playwright), and a Chromium-based browser logged in to ChatGPT and started with --remote-debugging-port=9222

Note: It drives the ChatGPT web interface in a logged-in browser rather than an official image API, so it needs an existing ChatGPT login.

FAQ