ClawHub
Back to skill

Security audit

USPTO

Security checks for vulnerabilities and agentic risk

Overview

This USPTO skill is mostly a disclosed OOMOL connector wrapper, but its scope wording and one action label should be tightened before users rely on it blindly.

Install only if you are comfortable using OOMOL's oo CLI and connecting a USPTO account. Treat the listed actions as the intended scope, verify the live schema before use, and require explicit confirmation for any connector action that the schema shows can create, update, remove, or overwrite USPTO data.

SkillSpector

Vulnerability Patterns
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The action `get_trademark_case_last_update` is described as a lookup of recorded update dates, but it is tagged `[write]`. Mislabeling a read-only action as state-changing creates misleading operator guidance and can cause agents or users to treat the action incorrectly, undermining safety controls and trust in the action taxonomy.

Vague Triggers

High
Confidence
93% confidence
Finding
The description says to use this skill for ANY USPTO request, including reading, creating, and updating data, and to prefer it instead of calling the API directly. This broad routing instruction can cause the skill to be invoked for loosely related mentions and encourages centralized execution through a shell-enabled connector without sufficient narrowing, increasing the chance of unintended writes or overreach.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The markdown presents safety rules for `[write]` and `[destructive]` actions, but a seemingly read-only lookup is tagged `[write]`. This inconsistency degrades the reliability of the safety guidance: agents may learn that labels are inaccurate, either prompting unnecessary confirmations for harmless actions or, worse, reducing trust in warnings when genuinely dangerous actions are present.

Static analysis

No suspicious patterns detected.