ClawHub
Back to skill

Security audit

clawddocs组合包

Security checks for vulnerabilities and agentic risk

Overview

This automation bundle is broad and somewhat inconsistently documented, but the risky capabilities are mostly disclosed and fit its stated automation purpose.

Install only if you are comfortable giving this bundle file-writing and command-execution authority. Review commands before running them, use a test directory for batch processing, avoid sensitive folders, and provide API keys only when you understand which member skill or local service needs them.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (4)

Intent-Code Divergence

Medium
Confidence
94% confidence
Finding
The documentation states the bundle does not support custom scripts, yet the manifest declares exec and bash capabilities. This mismatch can mislead users and reviewers about the real execution surface, causing them to underestimate the risk of arbitrary command/script execution through bundled members.

Intent-Code Divergence

Medium
Confidence
92% confidence
Finding
The manifest claims no external API is required, but later instructions tell users to configure API keys and make authenticated API requests. This contradiction can cause unsafe credential handling, incorrect trust assumptions, and unreviewed network exposure because users may not realize the bundle involves authenticated service interaction.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The bundle advertises read/exec/write/edit/bash automation capabilities and multi-step workflow execution without clearly warning that it can change files or system state. In an automation skill, this increases the chance that users invoke destructive or irreversible actions without informed consent or proper sandboxing.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The example batch command writes output files for every CSV file but does not warn that existing files may be created or overwritten. Users may run the example in production or sensitive directories and unintentionally alter local data at scale.

Static analysis

No suspicious patterns detected.