AutoShorts for Openclaw

An AI-powered video pipeline that automatically transforms long-form content into viral, platform-ready short clips with human-in-the-loop approval.

mutonby
v2.0.3
May 10, 2026
0
720
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install autoshorts

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install autoshorts using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).

Prefer to download?

Get the raw skill files in a ZIP archive.

What is AutoShorts?

AutoShorts is a sophisticated daily AI pipeline designed to maximize content reach by repurposing long videos into high-engagement TikToks, Instagram Reels, and YouTube Shorts. Leveraging the multimodal capabilities of Gemini 3 Flash and the precision of Whisper transcription, it identifies viral moments, performs frame-accurate cuts, and overlays high-contrast hooks tailored for social media success.

This skill is built to integrate seamlessly with Openclaw Skills, allowing for a fully automated yet human-gated workflow. It manages the entire lifecycle from video picking and transcription to rendering with FFmpeg and scheduling via the Upload-Post API, ensuring your social media presence is consistent and data-driven without the manual editing grind.

AutoShorts Use Cases

  • Automatically extracting 20-60 second viral highlights from podcasts or long-form webinars.
  • Streamlining a daily content creation workflow for creators who need consistent social media presence.
  • Repurposing landscape video archives into vertical 9:16 content for modern mobile platforms.
  • Data-driven content optimization where an AI learning loop improves hook quality based on real engagement metrics.

How AutoShorts Works

  1. The skill picks a video from the input folder using a round-robin cycle to ensure all content is processed.
  2. Whisper generates a high-accuracy transcript with word-level timestamps to guide the cutting process.
  3. Gemini 3 Flash analyzes the video and transcript to select the most viral moments and write compelling hook text.
  4. FFmpeg extracts the clips and overlays a high-visibility 'pill' style hook at the start of each video.
  5. The AI agent performs a visual QA on rendered frames and presents a candidate table to the user for approval.
  6. For approved clips, the skill generates platform-specific metadata including SEO titles and optimized Instagram captions.
  7. Approved content is scheduled for publication across TikTok, Instagram, and YouTube via the Upload-Post API.

AutoShorts Setup

Prerequisites

Ensure you have ffmpeg and Python 3.11+ installed on your system.

Installation

cd ~/Documents/skill-autoshorts
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Configuration

Create a .env file in the skill directory with the following keys:

UPLOAD_POST_API_KEY=your_key_here
UPLOAD_POST_PROFILE=your_profile_name
GEMINI_API_KEY=your_google_key
INPUT_FOLDER=/path/to/videos
OUTPUT_FOLDER=/path/to/clips

AutoShorts Data Schema & Taxonomy

AutoShorts maintains an organized directory structure and state management system:

File/Folder Description
state/processed.json Tracks processed video hashes, cycle counts, and timestamps.
output/<slug>/ Contains transcript.json, clips.json, and rendered MP4 files for each video.
learnings/HOT.md Stores evolved rules and patterns learned from past engagement data.
learnings/post-history.jsonl Append-only log of every published clip and its associated Gemini metadata.

AutoShorts Advanced Features

  • Weekly Learning Loop: Automatically updates a HOT.md file to refine AI selection based on real-world view counts and engagement.
  • Multimodal Visual QA: The agent uses its own vision capabilities to verify hook placement and rendering before showing clips to the user.
  • TikTok Draft Mode: Sends videos directly to the TikTok app inbox for final manual touches before going live.
  • Round-Robin Scheduling: Intelligent cycle management that prevents content exhaustion while allowing for fresh clips from old videos.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*