ClawHub
Back to skill

Security audit

Claw Ctct Security Patrol

Security checks for vulnerabilities and agentic risk

Overview

This appears to be a real OpenClaw security audit skill, but its full and scheduled modes can repeatedly send host-identifying audit metadata to a CTCT remote service with incomplete disclosure.

Install only if you trust the publisher and want a broad local security audit. Prefer local-only mode for privacy-sensitive machines. Avoid the scheduled --push setup unless you are comfortable sending scan summaries, installed-skill metadata, MAC address, hostname, and a persistent agent ID to the CTCT service on every run; if enabled, review how to list and remove the OpenClaw cron job.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The skill is presented as a local security inspection tool, but the code supports uploading audit results to a remote endpoint when run with --push. That creates a capability mismatch: operators may expect purely local analysis, yet the script can exfiltrate environment-derived security telemetry off-host, including summarized findings and identifiers.

Context-Inappropriate Capability

Medium
Confidence
98% confidence
Finding
The script derives MAC address, hostname, and a persistent agent ID, then transmits them in signed requests. These are device-identifying attributes unrelated to simply generating a readable local audit report, and they increase privacy risk, fleet tracking capability, and blast radius if the remote service is compromised.

Context-Inappropriate Capability

Medium
Confidence
96% confidence
Finding
The code enumerates installed skills and can submit the full skill inventory to a remote assessment API. Installed component inventory is sensitive environmental metadata and goes beyond the stated purpose of producing a local system security audit report.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The trigger conditions are broad enough to match common phrases like '安全检查' or '系统安全', which raises the risk of accidental invocation. Because this skill can run shell commands, create files, and potentially send data remotely, unintended activation materially increases risk.

Static analysis

Detected: suspicious.dangerous_exec

Shell command execution detected (child_process).

Critical
Code
suspicious.dangerous_exec
Location
scripts/openclaw-hybrid-audit-changeway.js:165