sjht-cam-anno

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:23 PM UTC
• Version:
  1.0.0

Overview

This is a local security-camera video annotation helper; it has real privacy and overwrite risks, but the behavior is disclosed, purpose-aligned, and not deceptive.

Install only if you are authorized to process the camera footage. Run it on a dedicated video folder with a dedicated output directory, use trusted ffmpeg/ffprobe binaries, and review or redact generated frames, annotations, and dataset.jsonl before training, sharing, or storing them long-term.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (4)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 90% confidence
• Finding: The skill instructs the agent to read local files, write output datasets, and invoke shell commands, but it does not declare permissions or boundaries for those capabilities. In a skill that processes surveillance footage under paths like /root/hair-cam, this increases the risk of unintended access to sensitive files, uncontrolled filesystem writes, or overbroad command execution.

Vague Triggers

Medium

• Confidence: 83% confidence
• Finding: The activation text is broad enough to trigger on generic requests like data annotation or video annotation, which could cause the skill to engage outside its intended surveillance-video scope. Overbroad activation can route unrelated user tasks into a workflow that reads local files and writes outputs, raising the chance of unnecessary access to sensitive data.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The skill handles surveillance-camera videos, extracts frames, and creates structured annotations without any privacy or data-handling warning. Because these inputs may contain identifiable people, homes, or sensitive behavior, omission of retention, consent, redaction, and output-file guidance creates a meaningful privacy and compliance risk.

Missing User Warnings

Medium

• Confidence: 81% confidence
• Finding: The script writes extracted frames and a manifest into caller-specified output locations, and ffmpeg is invoked with `-y`, which silently overwrites existing files. In a data-labeling workflow over `/root/hair-cam`, this can destroy prior annotations or unrelated files if an unsafe output directory or symlinked path is used.

VirusTotal

66/66 vendors flagged this skill as clean.