ClawHub
Back to skill

Security audit

veritaset

Security checks for vulnerabilities and agentic risk

Overview

The skill’s plan-comparison features fit its purpose, with a privacy consideration around sharing usage and travel estimates.

Comparison inputs are sent to Veritaset and retained according to its [privacy policy](https://veritaset.com/privacy/), which the skill does not link. Share aggregate usage and coarse travel estimates; omit names, phone numbers, and full bills. Verify final prices with the operator.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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 (1)

The comparison workflow sends personal plan and usage parameters to an external service without a privacy disclosure.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The workflow directs the agent to transmit usage, current-plan pricing, and potentially travel parameters to an external service without instructing it to disclose that transfer or minimize the information sent. The URL-based fallback can also expose these parameters in service or intermediary request logs. This is a limited privacy weakness rather than evidence of malicious exfiltration: the service is directly relevant to the requested comparison, and the skill does not request identifiers, credentials, or full bills.

Content

No source excerpt is available for this finding.