ClawHub
Back to skill

Security audit

Overwatch

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only transparency skill that may make agents chatty, but it does not add code execution, data access, credentials, or hidden behavior.

Install this only if you want frequent progress updates from agents. Treat its status updates as subordinate to higher-priority privacy, confidentiality, and safety instructions, especially for sensitive work where revealing task state may be inappropriate.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
91% confidence
Finding
The skill declares itself 'always-active' and frames violations as failures, creating a global behavioral override that can affect unrelated tasks. While positioned as a UX/transparency feature, this broad activation condition can conflict with task-specific safety, privacy, or least-privilege expectations by forcing unsolicited disclosures and process chatter in every workflow.

Vague Triggers

Low
Confidence
82% confidence
Finding
The rule 'if a human would reasonably wonder' is subjective and leaves the model to infer when disclosure is required. Ambiguous triggers increase inconsistent behavior and can cause oversharing of internal actions or needless interruption, especially when combined with the skill's mandatory language and always-active status.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.