Photo Capture Skill

Capture a fresh webcam image on macOS, Windows, or Linux. Prefer direct camera capture via ffmpeg so the workflow does not depend on screen-recording, access...

Security Scan

The skill's code, instructions, and requirements are coherent with its stated purpose (capturing webcam images locally via ffmpeg with a macOS screenshot fallback); it does not request unrelated credentials or contact external endpoints.

Purpose & Capability

Name/description match the actual code and runtime instructions: the scripts invoke ffmpeg (or macOS UI automation) to capture camera frames. No unrelated binaries, environment variables, or cloud credentials are requested.

Instruction Scope

SKILL.md directs the agent to run the included scripts from the skill directory and save images to tmp/; it also documents the macOS screenshot fallback which requires Accessibility/Screen Recording/Automation permissions. The instructions do not ask for unrelated files, credentials, or network exfiltration. Note: running the skill will normally trigger OS camera permission prompts for the terminal/Python process.

Install Mechanism

No install spec — instruction-only with bundled scripts. This is low-risk as no external archives are downloaded or installed by the skill itself. The only external dependency is ffmpeg, which the scripts expect to be on PATH.

Credentials

The skill requires no environment variables, credentials, or config paths. Access needs (camera device files like /dev/video* on Linux or OS permission dialogs) are proportional to a webcam-capture utility.

Persistence & Privilege

The skill is not always-enabled and does not request elevated platform persistence. It does not modify other skills or system-wide agent settings. Autonomous invocation is allowed by default (platform normal), but that is not combined with other concerning privileges here.

Assessment

This skill appears to do exactly what it says: capture a local webcam frame using ffmpeg or (on macOS fallback) by automating a camera app and taking a screenshot. Before installing or running it: ensure you trust the skill source (it will invoke local camera hardware and may trigger OS permission prompts), have ffmpeg installed and on PATH, and be aware the macOS fallback requires Accessibility/Screen Recording/Automation permissions. If you want extra safety, review the included scripts yourself or run the skill in a sandbox/VM; disable autonomous invocation if you do not want the agent to call it without your explicit request.

Photo Capture

Use this skill when the user wants a live camera image captured and sent back.

Default Behavior

1. Prefer direct webcam capture with scripts/capture_webcam.py.
2. Save output under tmp/ inside the workspace.
3. Reply with MEDIA:./relative/path.png or MEDIA:./relative/path.jpg so the image is sent back.
4. Only use the old macOS app-window screenshot path when direct camera capture is unavailable.

Why This Path

Direct camera capture is the default because it works across macOS, Windows, and Linux and does not require pre-configuring screen recording, accessibility, or automation permissions.

The OS may still show a normal one-time camera permission prompt for the terminal or Python process. That is expected and lighter than the old screenshot-based setup.

Primary Command Path

Note: Replace <skill-dir> with the actual path where this skill is installed (e.g., ~/.agents/skills/photo-capture or ~/.openclaw/workspace/skills/photo-capture).

List available cameras:

python3 <skill-dir>/scripts/capture_webcam.py --list-devices

Capture from the default camera on any supported OS:

python3 <skill-dir>/scripts/capture_webcam.py --output ./tmp/webcam.jpg

Capture from a specific device:

python3 <skill-dir>/scripts/capture_webcam.py \
  --device "0" \
  --output ./tmp/webcam.jpg

Higher resolution capture:

python3 <skill-dir>/scripts/capture_webcam.py \
  --width 1920 \
  --height 1080 \
  --output ./tmp/webcam-1080p.jpg

Platform Notes

• macOS: Uses ffmpeg with avfoundation. Auto-selects camera index 0 by default.
• Windows: Uses ffmpeg with dshow. Auto-selects the first detected video device.
• Linux: Uses ffmpeg with v4l2. Auto-selects /dev/video0-style devices.

Requirements

• ffmpeg must be available in PATH.
• A webcam device must be present.
• If multiple cameras exist, use --list-devices and then pass --device explicitly.

macOS Fallback

Keep the old app-window screenshot path only as a fallback for macOS environments where direct webcam capture fails.

Photo Booth fallback:

bash <skill-dir>/scripts/capture_via_app.sh \
  --app "Photo Booth" \
  --layout large \
  --capture window \
  --output ./tmp/photobooth-window-large.png

Use that fallback only if the direct ffmpeg path fails or the user explicitly wants the Photo Booth or FaceTime UI visible in the image.

Validation

When updating or debugging this skill:

1. Run --list-devices on the target OS.
2. Capture one frame to tmp/.
3. Confirm the output file exists and is not empty.
4. On macOS, only reach for the screenshot fallback if direct capture is blocked.