Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Webcam Motion Monitor
v1.0.3Webcam motion detection and monitoring system for WSL2 with USB/IP passthrough. Use when setting up motion detection on a USB webcam, monitoring camera snaps...
⭐ 0· 119·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Files and instructions match the stated purpose (motion detector, watcher, web preview, cleanup). Minor mismatches: snapshot.sh invokes motion_detector.py (nonexistent) while the actual detector is motion_detector_headless.py; web_preview.py docstring says port 8080 but it actually binds 8081. These look like sloppy/untested packaging rather than purposeful deception, but they will break expected usage.
Instruction Scope
SKILL.md instructs only camera-related actions and local file operations (create snapshots, logs, analysis queue). It also instructs the agent to analyze queued images with the 'image tool' and 'qwen model' but provides no details on what credentials or external services (if any) are required for that analysis. The watcher writes queue files and includes a hard-coded log string 'Person: Jade OR Sarah...' and the SKILL.md encourages storing person descriptions in MEMORY.md — this raises privacy concerns (sensitive biometric/identity info) but stays within the skill's stated domain.
Install Mechanism
No install spec; instruction-only plus bundled scripts. No network downloads or package installs declared. That reduces supply-chain risk, but also means the bundle's code will be written to disk if installed manually.
Credentials
No environment variables, credentials, or external config paths are requested. The scripts operate on user-owned paths under ~/.openclaw/workspace/camera which is proportionate to the stated purpose.
Persistence & Privilege
always is false and the skill does not request special system privileges, which is appropriate. However web_preview.py binds the HTTP server to 0.0.0.0 (not loopback), potentially exposing the camera stream beyond the local machine/WSL instance depending on network/WSL host mappings and firewall rules — this is a privacy/exposure risk. The watcher also writes queue files that an agent could read and analyze; autonomous invocation is allowed by default (not a direct flaw, but combined with queued images it increases blast radius).
What to consider before installing
What to check before installing or running this skill:
- Do not run it blindly as root. Run as a normal user and inspect the files yourself.
- Fixable bugs: snapshot.sh points to motion_detector.py which doesn't exist, and motion detector script lacks a --snapshot CLI; the web preview docstring references port 8080 but server uses 8081. Expect to edit/fix these if you rely on the helper scripts.
- Network exposure: web_preview.py binds the HTTP server to 0.0.0.0. If you only want localhost access, change server address to ('127.0.0.1', 8081) or ensure your host firewall prevents external access. Test reachability from other hosts before trusting privacy.
- Privacy: the skill encourages storing descriptive person-identification data (MEMORY.md) and queues images for AI analysis. Consider whether you have consent to capture and analyze people on camera. Review and limit retention (MAX_AGE_HOURS / CLEANUP_INTERVAL) and location of saved images.
- Image analysis ambiguity: SKILL.md mentions using an 'image tool' and the 'qwen' model but provides no details about where that runs (local vs external) or required credentials. Confirm how analysis will be performed and whether any external API calls or credentials are needed before enabling automatic analysis.
- Run initially in a controlled environment (air-gapped VM or isolated WSL instance) to validate behavior and to patch the small inconsistencies. If you accept those issues and restrict network access to the web server, the code otherwise appears to implement the declared functionality rather than exfiltrate data.Like a lobster shell, security has layers — review code before you run it.
latestvk97d1hkhkh6qr01dnrwntfcmdx8331g7
License
MIT-0
Free to use, modify, and redistribute. No attribution required.