AgentSkillsCN

replicate-studio

通过 Replicate API 实现 AI 内容生成。利用 Flux、SDXL、MusicGen 等模型，轻松创作图像、生成音乐、对照片进行超分辨率处理，或完成音频转录。适用于用户需要 AI 内容生成、图像创作、音频处理或模型推理的场景。

SKILL.md
--- frontmatter
name: replicate-studio
description: AI generation with Replicate API. Create images, music, upscale photos, transcribe audio using models like Flux, SDXL, MusicGen. Use when user needs AI content generation, image creation, audio processing, or model inference.
version: 1.0.0
author: Agent Zero Custom
tags: [ai, generation, images, audio, music, upscaling, transcription, replicate]
trigger_patterns:
  - "generate image"
  - "create image"
  - "AI image"
  - "upscale photo"
  - "transcribe audio"
  - "generate music"
  - "create music"
  - "text to image"
allowed_tools:
  - code_execution_tool
  - memory_save
  - response

Replicate Studio — AI Generation

Generate AI content using Replicate API. Supports image generation, audio processing, upscaling, and transcription.

Installation

bash
# Install dependencies
pip install -r requirements.txt

# Or use setup script
bash /a0/usr/skills/setup.sh

# Set API token (get at https://replicate.com)
export REPLICATE_API_TOKEN="your_token_here"
# Or add to /a0/.env: REPLICATE_API_TOKEN=your_token_here

When to Use

Use this skill when you need to:

  • •Generate images from text prompts
  • •Upscale low-resolution images
  • •Create music or audio
  • •Transcribe speech to text
  • •Run AI model inference

Supported Models

ModelTypeBest For
flux-proImageHigh-quality images
flux-schnellImageFast image generation
sdxlImageDetailed images
musicgenAudioMusic generation
esrganImage4x image upscaling
whisperAudioSpeech transcription

Usage

Via Python Script

bash
python /a0/usr/skills/replicate-studio/scripts/replicate_studio.py --model flux-pro --prompt "a beautiful sunset" --output /a0/tmp/sunset.png

Parameters

ParameterTypeDefaultDescription
--modelstrrequiredModel to use (flux-pro/flux-schnell/sdxl/musicgen/esrgan/whisper)
--promptstrrequiredInput prompt or description
--outputstrrequiredOutput file path
--input_imagestroptionalInput image for upscaling
--input_audiostroptionalInput audio for transcription

Examples

  1. •

    Generate image:

    bash
    python /a0/usr/skills/replicate-studio/scripts/replicate_studio.py --model flux-pro --prompt "futuristic city at night, neon lights, cyberpunk style" --output /a0/tmp/city.png
  2. •

    Upscale image:

    bash
    python /a0/usr/skills/replicate-studio/scripts/replicate_studio.py --model esrgan --input_image /a0/tmp/photo.png --output /a0/tmp/photo_4x.png
  3. •

    Transcribe audio:

    bash
    python /a0/usr/skills/replicate-studio/scripts/replicate_studio.py --model whisper --input_audio /a0/tmp/recording.mp3 --output /a0/tmp/transcription.txt

Requirements

  • •REPLICATE_API_TOKEN must be set in environment or /a0/.env file
  • •replicate>=0.22.0
  • •requests>=2.31.0

Files

code
/a0/usr/skills/replicate-studio/
├── scripts/
│   └── replicate_studio.py
├── requirements.txt
└── SKILL.md