ClawHub
Back to skill

Security audit

Budget Vs Actual

Security checks for vulnerabilities and agentic risk

Overview

This is a finance analysis skill that is coherent with its stated purpose and does not show hidden execution, persistence, or data exfiltration behavior.

Installers should treat this as a finance workflow helper and provide only the financial files or QBO access needed for the specific analysis. If no budget exists, use a separate budgeting workflow first or ask for explicit assumptions before relying on this skill's outputs.

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
95% confidence
Finding

Line L012 states the tester 'Constructed a realistic 2-month budget from scratch' while exercising the budget-vs-actual workflow, even though the skill manifest explicitly says this skill is 'NOT for building budgets from scratch.' This is a direct contradiction between the documented intended use in the test report and the stated skill intent.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

L012 claims the tester 'ran the full 7-step workflow as specified in SKILL.md' after constructing a budget from scratch, but later L049-L056 says the skill both forbids budget building and lacks guidance for cases with no budget. That makes the report's description of following the specified workflow internally inconsistent about whether this activity was actually supported by the skill.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.