ClawHub
Back to skill

Security audit

Probabilistic Thinking

Security checks across malware telemetry and agentic risk

Overview

This is a markdown-only reasoning skill for making calibrated probability estimates, with no executable code or hidden access requests.

Install if you want an agent to help structure uncertain judgments as probability estimates. Be mindful that it may activate in ordinary forecasting discussions, and avoid recording sensitive business, medical, legal, or personal details in calibration logs unless you intentionally choose where that information is stored.

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 (1)

Vague Triggers

Medium
Confidence
88% confidence
Finding
The manifest starts with specific trigger phrases, but then expands activation to broad conditions like 'user is making a forecast' or 'someone claims "I'm 90% sure"' and 'team is treating an uncertain outcome as binary.' These conditions are common in many conversations and do not clearly bound when the skill should activate, increasing the chance of over-triggering despite the provided negative examples.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.