Back to Skills
    🦞

    aimlapi-media-gen

    Generate images or videos via AIMLAPI

    By @d1m7asis
    View on GitHub
    SKILL.md
    ---
    name: aimlapi-media-gen
    description: Generate images or videos via AIMLAPI (OpenAI-compatible) from prompts. Use when Codex needs to call AI/ML API media endpoints, batch image/video jobs, or produce media files from text prompts using AIMLAPI credentials.
    ---
    
    # AIMLAPI Media Generation
    
    ## Overview
    
    Generate images and videos via the AIMLAPI OpenAI-compatible API using reusable scripts that handle auth, payloads, and media downloads.
    
    ## Quick start
    
    ```bash
    export AIMLAPI_API_KEY="sk-aimlapi-..."
    python3 {baseDir}/scripts/gen_image.py --prompt "ultra-detailed studio photo of a lobster astronaut"
    python3 {baseDir}/scripts/gen_video.py --prompt "slow drone shot of a foggy forest" --model aimlapi/<provider>/<video-model>
    ```
    
    ## Tasks
    
    ### Generate images
    
    Use `scripts/gen_image.py` to call `/images/generations` and save results locally.
    
    ```bash
    python3 {baseDir}/scripts/gen_image.py \
      --prompt "cozy cabin in a snowy forest" \
      --model aimlapi/openai/gpt-image-1 \
      --size 1024x1024 \
      --count 2 \
      --out-dir ./out/images
    ```
    
    **Notes:**
    
    - Change `--base-url` if your AIMLAPI endpoint differs.
    - Use `--extra-json` to pass provider-specific parameters without editing the script.
    
    ### Generate videos
    
    Use `scripts/gen_video.py` to call the video generation endpoint and save the returned media.
    
    ```bash
    python3 {baseDir}/scripts/gen_video.py \
      --prompt "time-lapse of clouds over a mountain range" \
      --model aimlapi/<provider>/<video-model> \
      --endpoint videos/generations \
      --out-dir ./out/videos \
      --extra-json '{"duration": 6, "size": "1024x576"}'
    ```
    
    **Notes:**
    
    - The video endpoint and payload fields vary by provider. Confirm the correct endpoint and fields in your AIMLAPI model docs and pass them via `--endpoint` and `--extra-json`.
    
    ## References
    
    - `references/aimlapi-media.md`: endpoint and payload notes, plus troubleshooting tips.