Camera

Capture photos from MacBook webcams. Use when user asks to take a photo, picture, snapshot, or see them. Two cameras available - Brio (front-facing on monitor) and FaceTime (side angle from MacBook).

Security Scan

The SKILL.md provides coherent ffmpeg commands to capture Mac webcams (so the functionality matches the description), but it omits declaring the required ffmpeg binary and allows model invocation of a sensitive capability (camera access) without mitigations — a privacy risk and an incoherence in declared requirements.

Purpose & Capability

The skill's purpose (take photos from Mac webcams) matches the runtime instructions (ffmpeg avfoundation commands using device indices). However, the skill does not declare that ffmpeg is required (required binaries list is empty), which is an inconsistency: the commands will fail unless ffmpeg is present. The SKILL.md's reliance on specific camera indices is plausible but platform-specific and should be declared.

Instruction Scope

Instructions are specific and confined to capturing frames and storing them under /tmp; they do not instruct reading unrelated files, exporting images to remote endpoints, or accessing environment variables. They do instruct running shell ffmpeg commands (including warmup videos) and saving outputs to /tmp, which is expected for this purpose. The SKILL.md does not include any explicit user-consent step before capture.

Install Mechanism

There is no install spec and no code files (instruction-only), which minimizes disk-write/remote-download risk. However, the skill implicitly depends on an external binary (ffmpeg) but does not list it as required or provide an install mechanism — an operational incoherence (not a direct remote-code risk, but it may cause unexpected failures).

Credentials

The skill requests no environment variables, credentials, or config paths — appropriate for a local camera capture utility. Nothing appears to ask for unrelated secrets or broad system config access.

Persistence & Privilege

The skill does not set always:true (so it's not force-included), but disable-model-invocation is not set, which allows the model to invoke the skill autonomously. Because the skill enables taking pictures from a webcam (a highly sensitive capability), allowing autonomous model invocation without an explicit user-confirmation step is a privacy concern. The skill also does not declare user-invocable or require explicit user consent flows.

What to consider before installing

This skill appears to do what it claims (use ffmpeg to grab webcam photos), but there are a few things to consider before installing: - Privacy: Camera access is sensitive. The skill allows model invocation by default (disable-model-invocation not set), meaning the agent could capture images without an explicit confirmation step. Prefer skills that require user-invocable-only or set disable-model-invocation: true for camera functionality. - Missing binary declaration: The SKILL.md runs ffmpeg but the registry metadata does not declare ffmpeg as a required binary or provide an install method. Confirm ffmpeg is installed and in PATH on your machine before use. - Platform & permissions: The commands use avfoundation device indices (macOS-specific) and will trigger OS camera-permission prompts. Make sure you trust the skill and inspect any prompts; the skill does not document a user-consent step. - Operational safety: Captured images are written to /tmp. If you install/use this, verify how and when images are uploaded or removed — the skill does not instruct uploading images but an agent could be instructed elsewhere to transmit them. Recommendations: only enable this skill if you trust its source; ask the publisher to (1) explicitly list ffmpeg as a required binary or provide an install spec, (2) mark it user-invocable and/or set disable-model-invocation: true (so captures require explicit user request), and (3) add an explicit confirm-before-capture step in SKILL.md. If you have low tolerance for privacy risk, do not enable it until these changes are made.

Camera Skill

Available Cameras

Camera	Index	Position	Best For
Brio 100	0	On external monitor, facing user directly	Front view, face shots
FaceTime HD	1	MacBook on right side, angled toward user	Side/profile view

Capture Commands

Use -loglevel error to suppress ffmpeg spam. Always warm up for 5s (camera needs exposure adjustment).

Brio (front view)

ffmpeg -loglevel error -f avfoundation -framerate 30 -i "0" -t 5 -y /tmp/brio_warmup.mp4 && \
ffmpeg -loglevel error -sseof -0.5 -i /tmp/brio_warmup.mp4 -frames:v 1 -update 1 -y /tmp/brio.jpg

FaceTime (side view)

Must use -pixel_format nv12 to avoid buffer errors.

ffmpeg -loglevel error -f avfoundation -pixel_format nv12 -framerate 30 -i "1" -t 5 -y /tmp/facetime_warmup.mp4 && \
ffmpeg -loglevel error -sseof -0.5 -i /tmp/facetime_warmup.mp4 -frames:v 1 -update 1 -y /tmp/facetime.jpg

Both cameras (parallel)

Run both commands simultaneously for multi-angle shots.

Output

• Photos saved to /tmp/brio.jpg and /tmp/facetime.jpg
• Warmup videos in /tmp/*_warmup.mp4 (can be deleted)
• Photos are ~80-100KB each

Gotchas

• Close Photo Booth or other camera apps first (can conflict)
• FaceTime camera REQUIRES -pixel_format nv12 or it fails with buffer errors
• 5s warmup is necessary for proper exposure