What problem does it solve? It lets you generate and edit images with GPT Image 2 (ChatGPT Images 2.0) directly inside your agent workflow, reusing your existing ChatGPT Plus or Pro subscription through the local Codex CLI instead of paying for separate OpenAI API access or per-image billing. ## Core Features & Use Cases - Text-to-image generation: Run a single bash script with a prompt and output path to produce PNG, JPEG, or WebP images decoded from the Codex session rollout. - Image-to-image editing and style transfer: Pass one or more reference images with --ref to repaint, restyle, or compose multi-reference outputs. - Use Case: You attach a flat-color icon and ask to repaint it as a 1950s ukiyo-e woodblock print; the skill runs codex exec with the image_generation feature enabled, extracts the base64 image payload from the new session file, and writes the result to your chosen path. ## Quick Start Ask the agent to use GPT Image 2 to generate an image from your prompt and save it to a specified output path.