ClawHub
Back to skill

Security audit

Finance Audit Agent

Security checks for vulnerabilities and agentic risk

Overview

This is a finance audit guidance skill that openly describes sensitive document review workflows and includes consent, minimization, no-exfiltration, and human-review limits.

Install only if you intend to use it for finance audit work involving regulated data. Users should confirm the module before sharing materials, provide desensitized documents where possible, and have qualified audit, compliance, or legal staff review outputs before relying on them.

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
  • 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 (2)

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill states that before first-round user confirmation it will not read, receive, or process materials containing personal or account information, but its examples and described workflows presume direct ingestion and parsing of IDs, bank cards, invoices, medical records, and account-related datasets. This inconsistency can cause unauthorized or premature handling of regulated data if upstream systems auto-route files or users assume the confirmation gate is enforced when it is only documented.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill claims it only provides methods, checklists, and report drafts, yet the rest of the document advertises OCR parsing, rule execution, data comparison, screening, and generation of concrete audit outputs. This mismatch can mislead operators, reviewers, or routing systems about the skill’s real data-handling and action-taking behavior, weakening governance controls and increasing the chance that sensitive financial or personal data is processed under false assumptions.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.