Security audit

Tax Deduction Optimizer

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed tax-deduction planning helper with no hidden execution, network use, credential handling, or persistence.

Install only if you want general tax-deduction planning help. Verify any deduction, filing, or timing strategy against the relevant jurisdiction and tax year with a qualified tax professional before acting.

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
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

Findings (8)

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The README promotes tax-minimization and deduction-conversion strategies across multiple jurisdictions (U.S. and China) without warning that tax rules are highly jurisdiction-specific, time-sensitive, and dependent on individual facts. Users could rely on the guidance as authoritative and take noncompliant positions, especially where the text suggests turning unreimbursed employee expenses into deductions or negotiating around tax bracket thresholds.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The trigger conditions are broad enough to activate on generic financial-planning requests, which can cause the skill to be invoked outside its intended tax-deduction-planning scope. In a tax context, over-activation increases the chance of irrelevant or overly specific tax guidance being presented without proper jurisdiction, qualification, or caution, which can mislead users making financial decisions.

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The module docstring says this is a 'Descriptive skill providing templates and frameworks,' which implies a generic template-oriented helper. In reality, the code analyzes user financial/tax-related text and produces targeted tax deduction recommendations such as home office, vehicle, and education deductions.

Natural-Language Policy Violations

Low

Confidence: 94% confidence
Finding: The acceptance criteria state 'Clear documentation in English,' which imposes a language requirement in natural-language policy without offering user opt-in or explaining a justified locale constraint. This matches the policy-violation category for forced language or locale requirements.

Natural-Language Policy Violations

Low

Confidence: 84% confidence
Finding: The file presents core skill behavior and examples in both Chinese and English, but does not state whether the skill adapts to the user's preferred language or locale. Under the policy, forcing or assuming a language/locale without opt-in can be a natural-language policy issue.

Natural-Language Policy Violations

Low

Confidence: 88% confidence
Finding: Scenario 4 is written entirely in Chinese and presents expected output in Chinese, but the document does not indicate that the skill supports multiple languages or that this locale-specific content is intentional and constrained. This can conflict with language/locale policy expectations when no user opt-in or explicit multilingual behavior is described.

Intent-Code Divergence

Low

Confidence: 92% confidence
Finding: The comment suggests the tax-deduction-specific logic is not yet implemented, but the subsequent lines immediately implement home office, vehicle, and education deduction recommendations. This actively misstates the code's current behavior and intent.

Natural-Language Policy Violations

Low

Confidence: 97% confidence
Finding: The manifest sets the skill language to "en", which imposes a specific language choice in natural-language metadata. Under the policy, locale or language restrictions should either be optional for the user or clearly justified as region-specific; no such opt-in or justification is present here.

Static analysis

No suspicious patterns detected.