Back to skill

Security audit

Southeast Asia VAT Pre-check

Security checks for vulnerabilities and agentic risk

Overview

This skill provides a limited tax pre-check using information you supply, though uploaded documents need careful redaction.

Prefer aggregate sales figures. Before uploading documents, remove tax identifiers, bank details, customer information, credentials, and unrelated transactions. Check your agent provider's upload and retention practices, and treat the report as preliminary rather than proof of tax compliance.

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
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)

Requests uploads of tax-filing data and platform statements without a privacy warning or redaction guidance.

Medium
Category
Output Handling
Confidence
87% confidence
Finding

The skill invites uploads of draft tax filings and platform statements without advising users to remove tax identifiers, customer details, banking information, or other unnecessary sensitive data. Although transaction information is relevant to the pre-check, full documents can expose substantially more information than the aggregate figures requested in Step 1. The prohibition on requesting credentials or network access reduces risk, but does not prevent sensitive information from being unnecessarily submitted to the agent.

Content

No source excerpt is available for this finding.