ClawHub
Back to skill

Security audit

Python基础版

Security checks for vulnerabilities and agentic risk

Overview

This skill is mainly a Python review helper, but it asks for and describes broader command, API, and file-writing abilities that are not clearly bounded for that purpose.

Install only if you are comfortable with a Python review skill being selected for broad development tasks and potentially using command execution or network/API workflows. Prefer a revised version that limits tools to read-only review, narrows the trigger text, and specifies any allowed commands or external calls explicitly.

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

High
Confidence
95% confidence
Finding
The manifest presents this as a narrow Python code-review skill, but the body expands it into broad automation functions such as file handling, API integration, and command execution. This capability drift can cause an agent or operator to grant or invoke privileges far beyond the user's expected scope, increasing the chance of unsafe actions and misuse.

Context-Inappropriate Capability

Medium
Confidence
89% confidence
Finding
Introducing network/API access for a basic Python quality-review skill expands the attack surface without a clear functional need. Even if no direct credential is requested here, network capability can enable data exfiltration, unintended external calls, or confused-deputy behavior when the skill is invoked on sensitive code.

Context-Inappropriate Capability

High
Confidence
97% confidence
Finding
Documenting system command execution for a basic code-review skill is dangerous because command execution is a high-risk primitive unrelated to simple static guidance. In the skill context, this can normalize running shell commands based on user-supplied code or troubleshooting instructions, creating a path to destructive local actions or environment reconnaissance.

Vague Triggers

Medium
Confidence
83% confidence
Finding
The activation language is broad enough to match general development, debugging, deployment, and coding-assistance requests, not just the narrow Python hygiene tasks described. Over-broad matching can cause this skill to be selected in contexts where its permissions or behavior are inappropriate, especially given the documented drift toward execution and automation features.

Static analysis

No suspicious patterns detected.