ClawHub
Back to skill

Security audit

AI Agent Governance（AI智能体治理）

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AI-agent governance guide with a small local Python helper that prints assessments and does not show hidden network, credential, persistence, or destructive behavior.

Reasonable to install if you want an offline AI-agent governance reference and checklist tool. Run the Python helper only when you intend to execute local code, and independently verify legal/regulatory claims before relying on them for compliance decisions.

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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)

Lp3

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding
The skill advertises executable local commands and shell usage (`python tools/agent_gov_toolkit.py ...`) but does not declare any permissions despite requiring code-capable operations such as shell execution and likely file access. This creates a governance gap: users and the platform may not realize the skill can invoke local tooling, reducing transparency and weakening review, sandboxing, and least-privilege enforcement.

Natural-Language Policy Violations

Medium
Confidence
94% confidence
Finding
The file explicitly mandates a single branding form ('注册老炮') and forbids use of personal name, employer, or internal path in outward-facing releases. In an agent skill context, hard-coded authorial/branding constraints can override user or organizational preferences, create deceptive attribution, and pressure downstream agents to emit fixed identity text without user opt-in.

Static analysis

No suspicious patterns detected.