Skip to content
TakoAPI
Back to skills

sogni-gen

by krunkosaurus

Coding Agents & IDEs
622Downloads reported by clawskills.sh
2Stars
0Views
0 likesSign in to like this skill
—Not yet rated
Your rating

Brief

Generate images **and videos** using Sogni AI's decentralized network, with local credential/config files.

Description

What This Skill Does

Connects to Sogni AI's decentralized GPU network to generate images and videos from text prompts or reference files. Supports editing existing images, face transfer via photobooth mode, 360-degree turntable sweeps, and multiple video workflows including text-to-video, image-to-video, audio-driven, and video-to-video with ControlNet.

Uses a decentralized GPU network instead of a central API endpoint, which can reduce queuing and avoids single-provider lock-in for compute.

When to Use It

  • Generate a stylized portrait from a face photo for a LinkedIn headshot
  • Animate a product image into a short looping video
  • Restore a damaged vintage photo by editing it with a text prompt
  • Create a 360-degree turntable video from a single product image
  • Generate a lip-synced talking head from a face image and audio file

Example Workflow

INPUT

User asks: generate a short video of ocean waves at sunset

AGENT
  1. 1

    Selects text-to-video workflow with default WAN model

  2. 2

    Rewrites the prompt into a continuous cinematic prose paragraph

  3. 3

    Runs sogni-gen.mjs with --video flag and -o flag to save locally

  4. 4

    Receives JSON output with success status and local file path

OUTPUT

An mp4 file saved to the specified output path, e.g. /tmp/video.mp4

Requirements

  • Sogni AI account at https://app.sogni.ai/ (free tier available with daily Spark points)
  • SOGNI_API_KEY environment variable or SOGNI_USERNAME and SOGNI_PASSWORD credentials
  • ffmpeg installed (required for 360-degree video assembly and video concatenation)

Installation

clawhub install krunkosaurus/sogni-gen

Run the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.

Links