ClawHub
Back to skill

Security audit

CSDN

Security checks across malware telemetry and agentic risk

Overview

This skill is a simple informational markdown file with no code or permissions, though its title and description do not match its actual guidance very well.

Safe to install from a security perspective, but review the content first: it may not provide the CSDN-focused career, blog, or open-source guidance promised by its description.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Description-Behavior Mismatch

Medium
Confidence
93% confidence
Finding
The manifest advertises a CSDN-focused skill for career development, technical blogs, and open-source guidance, but the body documents generic collaboration/software-selection criteria instead. This mismatch can mislead users and downstream agents about the skill’s actual purpose, causing incorrect invocation, unsafe trust assumptions, or inappropriate routing of user requests.

Intent-Code Divergence

Medium
Confidence
91% confidence
Finding
The skill claims to be about 'CSDN技术栈', but the listed filters, return fields, and examples describe office collaboration/productivity features unrelated to the stated CSDN intent. In agent ecosystems, contradictory documentation is dangerous because it can conceal true behavior, degrade policy enforcement, and make security review or user consent unreliable.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.