Installable agent tool for generating images with OpenAI's gpt-image-2 model (April 2026 — OpenAI's most capable image model). Wraps a Python CLI that handles API auth, parameter validation, file I/O, and cost estimation — the agent composes one command with a prompt plus optional --size / --quality / --format / --n / --output flags and receives back saved file paths and a JSON metadata block (in --json mode). For any image task — logo, app icon, hero banner, social post, photo-realistic render, draft test, multi-variation set — the agent infers suitable parameters from task context and produces the artifact in a single subprocess call, then hands the saved path(s) to the next step. ACTIVATE when an image is needed AND OpenAI is the chosen provider — user mentions "OpenAI", "gpt-image", "gpt-image-2", invokes /openai-image, or a parent task has explicitly routed an image step here. DO NOT auto-trigger on generic "generate an image" requests with no provider specified — defer to whichever provider the user pic
2026-05-29