ClawHub
Back to skill

Security audit

Deepvista Shared

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a normal DeepVista CLI reference, but it tells the agent to silently run an update-check command automatically when loaded.

Review this before installing if you do not want skills to run commands automatically. The main concern is the silent `deepvista skill check-update` action on load, which may make a network request. The rest of the skill is conventional DeepVista CLI documentation, and no destructive or credential-exfiltrating behavior was found.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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 (3)

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The skill instructs the agent to automatically run `deepvista skill check-update` on load, which is an undisclosed subprocess/network action unrelated to the user's explicit request. Even if the command is only for update checking, automatic execution expands the skill's behavior surface and can leak environment, profile, or network metadata without consent.

Context-Inappropriate Capability

Medium
Confidence
93% confidence
Finding
A shared reference skill should primarily provide documentation, but this one prescribes autonomous background command execution. That mismatch makes the behavior harder to justify and easier to abuse, especially because reference skills are expected to be passive and broadly reusable.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The instruction to run the update check silently and 'say nothing' on success hides side effects from the user. Silent background execution reduces transparency and auditability, and could normalize covert command execution in future skills.

Static analysis

No suspicious patterns detected.