ClawHub
Back to skill

Security audit

Code Analyze

Security checks for vulnerabilities and agentic risk

Overview

The skill is a plausible code-analysis helper, but it gives conflicting privacy guidance around local-only storage versus optional external LLM and webhook use.

Review this carefully before installing. Use it only on code you are allowed to analyze and share, disable or avoid external LLM endpoints and webhook/email notifications unless approved, and check the .code-analyze history and reports because they may retain sensitive findings locally.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The skill presents itself as locally/offline usable and stores history locally, but elsewhere documents optional external LLM API calls and webhook notifications. This creates a misleading data-flow model: users may assume sensitive code and analysis artifacts remain local when batch analysis or notifications could transmit them externally.

Intent-Code Divergence

High
Confidence
98% confidence
Finding
The statement that data 'does not leave the machine' directly conflicts with sections describing external API usage and webhook channels. This is dangerous because it can cause users to process sensitive source code, secrets, or architectural data under false privacy assumptions, leading to unauthorized disclosure.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The skill instructs the agent to create configuration, framework, report, and history files and to auto-archive analysis results, but it does not clearly warn users that local files will be created and retained. In a security-review context, those files may contain sensitive code findings, paths, and risk summaries that persist longer than the user expects.

Missing User Warnings

High
Confidence
97% confidence
Finding
The documentation describes optional external LLM API usage and API key configuration without clearly warning that source code, reports, or other user content may be sent to third-party services during batch analysis. In this skill's context—enterprise codebase and security review—that omission is especially dangerous because the processed material is likely sensitive.

Static analysis

No suspicious patterns detected.