ClawHub
Back to skill

Security audit

Norman: Categorize Transactions

Security checks for vulnerabilities and agentic risk

Overview

The skill is coherent for bookkeeping, but it can modify accounting records without clearly requiring user approval first.

Install only if you expect the agent to work inside your accounting system. Before using it, confirm the correct chart of accounts and require review and approval before categorizing transactions, linking invoices or receipts, or marking entries verified.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (3)

Vague Triggers

Medium
Confidence
94% confidence
Finding
The skill description is broad enough to match many ordinary finance or bookkeeping requests, which can cause the agent to invoke this skill in situations where the user did not clearly intend transaction edits or reconciliation actions. In this skill, overbroad activation is more dangerous because the workflow includes state-changing operations on accounting data, such as categorization, linking records, and verification.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The instructions direct the agent to modify accounting records by categorizing transactions, linking invoices and attachments, and marking transactions as verified, but they do not require an explicit warning or confirmation from the user. This is dangerous because users may treat the workflow as advisory while the agent performs irreversible or audit-relevant bookkeeping changes behind the scenes.

Natural-Language Policy Violations

Medium
Confidence
88% confidence
Finding
The skill hardcodes SKR04 bookkeeping categories for German businesses without checking whether the user's entity, jurisdiction, or chart of accounts actually matches that standard. In the accounting context, applying the wrong standard can misclassify entries, create compliance issues, and corrupt financial records at scale.

Static analysis

No suspicious patterns detected.