ClawHub
Back to skill

Security audit

Ham Radio DX Monitor

Security checks across malware telemetry and agentic risk

Overview

This skill does what it advertises: it connects to ham radio DX cluster servers, optionally sends a callsign, and supports user-configured local logging and cron monitoring.

Install this only if you want your agent to contact external DX cluster servers. Use a callsign only if you are comfortable sending it to the selected cluster. Add the cron entries only for intentional background monitoring, review or delete the /tmp logs and state file as needed, and do not run the missing setup script from any outside source unless you inspect it separately.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding
The skill documents commands that write to local files such as /tmp/dx-alerts.log, dated digest files in the user's home directory, and a state file at /tmp/dx-monitor-state.json, but it declares no permissions to reflect that behavior. This is not overtly malicious, yet undeclared write capability reduces transparency and can surprise users or bypass permission-based review expectations.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The example prompts are generic phrases like 'Check the DX cluster for new spots' and 'Show me today's DX digest,' which are broad enough to overlap with normal conversation and may trigger the skill unintentionally. Because this skill performs network connections to external servers and may transmit a callsign, accidental invocation has privacy and unintended-action implications.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The quick-start section shows connections to external DX cluster nodes and an option to send the user's callsign, but the description does not warn users that their client will contact third-party servers and may disclose identifying radio information. In context this is expected functionality, but the missing notice creates a meaningful consent and privacy gap.

Session Persistence

Medium
Category
Rogue Agent
Content
def load_state() -> Dict:
    """Load previous state."""
    try:
        with open(STATE_FILE) as f:
            return json.load(f)
Confidence
78% confidence
Finding
Load previous state

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.