toolguard-daemon-control

• Outcome: Review
• Latest audit: May 28, 2026, 1:24 PM
• Version:
  0.0.1

Overview

This skill is openly meant to manage macOS background services, but its installer has unsafe input handling and can immediately create persistent auto-restarting services.

Use only if you explicitly need a macOS launchd service manager. Before installing, review the exact command, arguments, working directory, and environment values; do not pass secrets through --env. Prefer a patched version that removes eval, validates service names and environment keys, escapes plist XML, and asks for confirmation before loading persistent services.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Rogue Agent: Self-Modification, Session Persistence
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (5)

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill explicitly creates persistent launchd agents, auto-restarts them, and writes plist and log files on disk, but it does not present a prominent user-facing warning about these durable side effects. This can lead to users unintentionally installing background persistence that survives the current session and continues running without clear consent.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The script writes a launchd plist and immediately loads it, creating a persistent background service without any confirmation or explicit warning to the user. In a daemon-management skill this behavior is expected, but it is still security-relevant because persistence is established as soon as the script is run, which can surprise users and make unintended long-lived execution easier.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: Environment variables provided with --env are embedded directly into the generated plist on disk in plaintext. If callers pass secrets such as API keys or tokens, those values become recoverable from the user's LaunchAgents directory and may also be exposed through backups or local inspection.

Session Persistence

Medium

• Category: Rogue Agent

Content

EOF

# Load and start service
launchctl load "$PLIST_PATH"

echo "Service '${SERVICE_NAME}' installed and started."
echo "  Label:  ${LABEL}"

• Confidence: 91% confidence
• Finding: launchctl load

Session Persistence

Medium

• Category: Rogue Agent

Content

EOF

# Load and start service
launchctl load "$PLIST_PATH"

echo "Service '${SERVICE_NAME}' installed and started."
echo "  Label:  ${LABEL}"

• Confidence: 91% confidence
• Finding: PLIST

VirusTotal

63/63 vendors flagged this skill as clean.