webcam-captureWebcam Capture
Give your AI assistant eyes. Capture photos and short videos from any connected webcam on demand. Hold something up to the camera — a whiteboard, document, product, or workspace — and ask the AI to analyze it. The camera light always blinks on capture, so you'll always know when it's active.
Key Features
- •Capture still photos from any connected webcam
- •Record short video clips (up to any duration)
- •Works with built-in, USB, or IP cameras
- •Privacy: camera light always blinks on capture
- •Specify which camera to use (for multi-camera setups)
- •Analyze captured images with AI vision models
Common Use Cases
- →Show the AI a whiteboard or diagram to explain
- →Document your physical workspace or desk setup
- →Capture receipts, documents, or products for records
- →Have AI analyze something you're holding up to the camera
- →Record a quick video of a demo or process
Custom Workflow Integration
This skill can be customized for your specific workflow as part of an SMF Works services engagement. Whether you need custom automation rules, integrations with your existing tools, or specialized configurations for your team, we can tailor this skill to fit your exact requirements.
Explore ServicesInstallation
# Install the skill (via TUI or CLI)
smfw install webcam-capture
# Get help
smfw run webcam-capture --help
💡 Tip: Install via the OpenClaw TUI skill manager for an interactive experience, or use the CLI command above.
Setup Guide
Webcam Capture — Setup Guide
Requirements
ffmpeginstalled on your system- A working webcam (built-in or USB)
- Linux/macOS/Windows (WSL works too)
Install ffmpeg
macOS:
brew install ffmpeg
Ubuntu/Debian:
sudo apt update && sudo apt install ffmpeg
Windows: Download from ffmpeg.org or use WSL.
Verify Your Camera
# List available cameras
ls /dev/video* # Linux
ls /dev/media* # Linux (more detailed)
# Test capture
ffmpeg -f avfoundation -list_devices true -i "" 2>&1 | grep -i camera # macOS
v4l2-ctl --list-devices # Linux
Set Permissions (Linux)
If you get permission errors:
# Check your user is in the video group
groups $USER
# Add yourself to video group
sudo usermod -a -G video $USER
# Then log out and back in
Quick Test
# Capture a test photo
ffmpeg -y -f v4l2 -input_format mjpeg -i /dev/video0 -frames:v 1 -update 1 test.jpg
# Or use the script directly
python3 capture.py --list-devices
How-To Guide
Webcam Capture — How-To Guide
Step 1: Capture a Photo
Open your OpenClaw TUI and say something like:
"Take a photo with the webcam"
The AI will run:
python3 ~/.openclaw/skills/webcam-capture/scripts/capture.py
Photos save to: ~/.openclaw/workspace/captures/
Or specify a custom path:
"Take a photo and save it to ~/Desktop/photo.jpg"
Step 2: Share What You're Looking At
Hold something up to the camera — a whiteboard, document, product, whatever — and say:
"Analyze what I'm showing you on the camera"
The AI captures a frame and describes what it sees.
Step 3: Record a Video Clip
"Record a 5-second video clip"
The AI runs:
python3 ~/.openclaw/skills/webcam-capture/scripts/capture.py --stream --duration 5
Videos save to: ~/.openclaw/workspace/captures/stream_YYYYMMDD_HHMMSS.mp4
Step 4: Use Multiple Cameras
If you have multiple cameras, tell the AI which one to use:
"Take a photo using the second camera"
The script supports --device /dev/video0 through /dev/videoN.
Step 5: Document Your Workspace
Great for documentation:
"Snap a photo of my workspace for my project notes"
The AI captures and can include the image in generated documentation.
Privacy
The camera light always blinks when capturing. No silent surveillance is possible — you'll always know when the camera is active.
Troubleshooting
| Problem | Solution |
|---|---|
Device not found | Run --list-devices to find your camera |
Permission denied | Add yourself to the video group: sudo usermod -a -G video $USER |
| Dark image | Your room may be too dark — the ring light helps! |
| Wrong camera | Specify device: --device /dev/video1 |
Script Reference
# Basic capture
python3 capture.py
# Custom output
python3 capture.py -o ~/my_photo.jpg
# Specific camera
python3 capture.py -d /dev/video1
# List cameras
python3 capture.py --list-devices
# Record video
python3 capture.py --stream --duration 10