grab

What This Skill Does

The grab skill is a powerful, locally-focused archiving tool designed to preserve content from the modern web. Instead of relying on volatile bookmarks or fleeting social media posts, grab downloads and archives content from platforms like X (Twitter), Reddit, and YouTube directly to your local file system. It captures raw media, text, transcripts, and metadata, ensuring your research is persistent and accessible offline. When configured with an OpenAI API key, the skill enhances these archives with AI-generated summaries and intelligent, descriptive folder naming. It leverages the local Whisper model for transcriptions, ensuring that high-quality text extraction from video and audio occurs entirely on your device without the need for additional network calls or privacy compromises.

Installation

To get started, you must ensure the required system dependencies are installed. Open your terminal and run:

brew install yt-dlp ffmpeg openai-whisper

Once the system dependencies are installed, you can install the skill via the OpenClaw manager:

clawhub install openclaw/skills/skills/jamesalmeida/grab

During the first run, the tool will prompt you for a save directory. This path is stored in ~/.config/grab/config and can be reset at any time by running grab --config.

Use Cases

• Research Archiving: Save long-form X articles and Reddit threads for offline reading and cross-referencing.
• Content Creation: Quickly batch-save YouTube videos and their associated transcripts to use for video editing, blog post generation, or AI training data.
• Social Monitoring: Capture tweets or Reddit discussions including all comments and metadata to track sentiment or community engagement over time.
• Personal Knowledge Management: Centralize digital content into a consistent folder structure that can be easily indexed by your favorite local search or knowledge base tools.

Example Prompts

1. "Grab this thread from Reddit and make sure to include the comments so I can analyze the sentiment later."
2. "Download the YouTube video at this URL, transcribe it, and summarize the key findings into my archives."
3. "Archive this X article and save it in my Dropbox; use the smart naming feature to label the folder based on the topic."

Tips & Limitations

• Efficiency: The transcription engine runs locally. While this protects your data, it may consume significant CPU resources during the initial processing of long videos.
• Configuration: Always ensure your OPENAI_API_KEY is set in your environment if you want the automated summaries. Without the key, the tool functions perfectly as a downloader, but you lose the generative insights.
• Browser Fallback: The tool is designed to intelligently switch to the OpenClaw browser snapshot method if it detects that an API (like Reddit's) is blocking direct access, ensuring high reliability even against anti-scraping measures.