OpenClaw adds WayinVideo AI Clipping skill for 10 Shorts from one YouTube link

• A new WayinVideo ClawHub skill lets OpenClaw turn a single YouTube link into multiple vertical clips with captions and reframing, as shown in the launch thread launch thread.
• Setup is lightweight: install WayinVideo/ai-clipping, then add a WAYIN_API_KEY, either by chatting with the agent or via the official skill docs; the same flow appears in setup tweet.
• The phone-first trick is the interesting part for creators: the demo routes clipping through Telegram, where OpenClaw watches the source video, picks moments, edits, and returns Shorts without a laptop Telegram workflow.
• WayinVideo says each rendered clip can include titles, descriptions, hashtags, a viral score, and a download link, while the thread shows the clips can also be pulled back into the WayinVideo dashboard for quick tweaks dashboard tweet.

You can install the skill from ClawHub, read the skill guide, and check the broader OpenClaw plus WayinVideo walkthrough. The docs also spell out a few useful specifics that the thread only hints at: clips default to 9:16, exports can be capped at 90 seconds, and billing runs at about 2 API Units per minute of input video.

The setup path is exactly what you want from an agent skill: install the package, paste in a WayinVideo API key, then hand the agent a link. The official docs say OpenClaw can install it conversationally with Install the WayinVideo/ai-clipping skill, and the same docs also give a manual CLI path with clawhub install WayinVideo/ai-clipping.

ClawHub lists the package as MIT-0 licensed, with Python 3 as the only runtime requirement and WAYIN_API_KEY as the only required credential. For a creator tool, that is refreshingly barebones.

The demo uses Telegram as the front end. Hasan Toxr shows a simple pattern: send a YouTube link in chat, let OpenClaw watch the source, find standout moments, add captions, reframe for vertical, and return post-ready Shorts.

That lines up with WayinVideo's own integration post, which pitches OpenClaw as a chat-native controller for video jobs across Telegram, WhatsApp, and Discord. Telegram gets the spotlight here because it makes the workflow feel oddly portable: clip first, open the laptop later.

The product surface is broader than "ten clips." According to the skill docs and the lower-level AI Clipping API docs, rendered outputs can include:

• vertical 9:16 framing by default
• clip durations up to 90 seconds for the standard social flow
• animated captions, including translated captions
• AI-generated titles, descriptions, and hashtags
• a virality-ranked clip list and direct download links

The thread adds one practical detail the docs do not emphasize as much: finished clips can be downloaded straight from Telegram or reopened inside the WayinVideo dashboard for reframing and quick edits dashboard tweet.

WayinVideo's ClawHub page exposes the operational details behind the slick demo. Direct URL processing works on a long list of platforms, including YouTube, Vimeo, Dailymotion, Kick, Twitch, TikTok, Facebook, Zoom, Rumble, and Google Drive, according to the skill package page and API docs. Unsupported sources have to be downloaded first and uploaded as files.

The package also writes JSON result files into a local workspace and uploads video data to WayinVideo's servers for processing, both called out in ClawHub's security review. That makes the skill feel less like a toy wrapper and more like a real production handoff between a chat agent, a video API, and a lightweight local runtime.