kangarooking/agnes-free-model-skills

This repository is a set of three agent skills that plug into Codex or similar AI agent tools, letting them call Agnes AI's free text, image, and video generation models. It only contains the reusable skill files, reference documentation, and small Python helper scripts. It does not include any API keys, generated content, or output files from running the scripts. The three skills cover different jobs. The text skill calls Agnes's free flash text model and supports normal chat responses, streaming output, and experimental tool calling. The image skill calls Agnes's free image model to generate a picture from a text description, or to transform an existing image, and downloads the result. The video skill calls Agnes's free video model, which works asynchronously: you start a video generation task, then poll it until it finishes and download the finished clip. To install a skill, you copy its folder into your Codex skills directory, or you can keep the whole repository anywhere and point your own agent runtime at the relevant SKILL.md file. Before using any of the scripts, you need an Agnes API key, which you provide through an environment variable rather than writing it into any file. The helper scripts also accept an alternate environment variable name, and you can override the API's base address if it ever changes. Each script can be run directly from the command line, for example asking the text model to explain something in three sentences, generating an illustration from a prompt, or creating a short vertical video clip. The project is explicit about its security boundaries: keys are read only from environment variables, no real keys or generated media are checked into the repository, and common output folders are excluded through gitignore. It is released under the MIT license.