ClawHub

V19 Sparse Policy Auditor

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only audit skill, but users should be careful because its examples send governance/audit data to an external service.

Install only if you trust the external governance endpoint. Use a dedicated limited key, do not paste secrets or private logs into the examples, and redact sensitive agent behavior, user data, and operational details before submitting audit evidence.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Context-Inappropriate Capability

Medium
Confidence
94% confidence
Finding
The skill is presented as a sparse policy auditor, but it also documents an unrelated agent registration workflow. Expanding scope in documentation can prompt users or agents to perform unnecessary external actions, increasing attack surface and enabling unintended enrollment or data submission to a third-party service.

Context-Inappropriate Capability

Medium
Confidence
97% confidence
Finding
Publishing a live health-check endpoint together with a shared governance key exceeds the stated audit purpose and provides reusable credentials to an external service. Even if intended for testing, embedded shared keys can be abused by anyone reading the skill to access, probe, or operationalize the service outside intended controls.

Missing User Warnings

Medium
Confidence
98% confidence
Finding
The markdown provides executable API examples that send audit context and evidence to an external endpoint, while also exposing a governance key and omitting warnings about credential handling, third-party transmission, retention, or trust boundaries. In an agent setting, this creates a strong risk of sensitive operational data being exfiltrated to an external service by default.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal