Back to skill

Security audit

Security App Wealth Advisor

Security checks for vulnerabilities and agentic risk

Overview

ClawScan’s verdict is benign with medium confidence: scanners reported one low-severity concern about product eligibility rules, while the available scan found no other reported findings; coverage was partial.

The verdict is benign, but confidence is medium and the scan was partial: Skillspector inspected part of one Markdown file, and the manifest issue and missing reference limited coverage. Before installing, consider whether the reported return-target exclusion fits your intended use, especially if you plan to rely on the skill for real product matching.

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
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

L264 documents below-target returns as a soft preference handled by score reduction, but L187 excludes products entirely when their expected return is more than two percentage points below the customer's target.

Low
Category
Excessive Agency
Confidence
97% confidence
Finding

The implementation treats returns more than two percentage points below the customer's target as a hard exclusion, contradicting the documented rule that return targets are soft preferences. If reused in a real matching system, an inflated return target could suppress otherwise suitable products and bias the candidate set toward higher projected returns, although the separate risk-band filter remains enforced. The explicitly illustrative scope and required human review reduce immediate exposure.

Content

No source excerpt is available for this finding.