Security audit

Tax Paperwork Staging Board

Security checks for vulnerabilities and agentic risk

Overview

This is a prompt-only tax paperwork organizer with no code or network access, but users should verify deadlines and tax treatment with official sources or a professional.

Install only if you want a prompt-based organizer for tax documents. Do not rely on it for filing deadlines, deductibility, eligibility, or legal/tax conclusions, and avoid giving it full taxpayer IDs, passwords, or account access.

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)

Natural-Language Policy Violations

Medium

Confidence: 97% confidence
Finding: The acceptance criteria require output to be 'English-first' and prohibit 'CJK-dominant paragraphs.' This is a natural-language locale policy constraint that forces a language preference without user opt-in or a documented region-specific justification.

Natural-Language Policy Violations

Medium

Confidence: 90% confidence
Finding: The README includes a full usage example in Chinese, but the document does not indicate that the skill is region-specific or that users can choose their preferred language. This can violate language/locale policy expectations by implicitly forcing a specific language experience for part of the skill description.

Intent-Code Divergence

Medium

Confidence: 96% confidence
Finding: The skill's safety section says it should not calculate tax-related outcomes or replace official guidance, yet Scenario 3 instructs it to create a new board with deadlines shifted by +1 year. Even if framed as administrative, changing deadlines is a date calculation that may be wrong across jurisdictions or filing contexts, creating a risk of missed compliance deadlines.

Description-Behavior Mismatch

Medium

Confidence: 94% confidence
Finding: The usage scenarios describe operational capabilities like moving items between board states, archiving a board, cloning templates, and shifting deadlines, which exceed the documented scope of simple document staging. This mismatch can cause downstream agents or users to over-trust the skill with workflow automation it was not designed or safety-bounded to perform, leading to unauthorized state changes or incorrect deadline handling.

Intent-Code Divergence

Medium

Confidence: 94% confidence
Finding: The Chinese scenario goes beyond neutral document staging by telling the user to track potentially deductible items and to file immediately using a tax app's prefill function. That crosses into procedural tax guidance and implies what records may support deductions, which conflicts with the stated prohibition on tax advice and may mislead users into acting without professional or official verification.

Natural-Language Policy Violations

Medium

Confidence: 91% confidence
Finding: Scenario 4 switches to Chinese and gives China-specific tax-filing guidance, but the skill does not state that it is multilingual or that Chinese output is optional and user-selected. This can violate language/locale policy expectations by implicitly forcing a language/locale mode without documented opt-in or justification.

Vague Triggers

Low

Confidence: 83% confidence
Finding: The activation examples include phrases like "Help me organize documents for my accountant" and "What should I put in my tax folder?", which are natural everyday requests and are not clearly constrained to this specific skill context. The file does not provide exclusion conditions or negative examples to clarify when the skill should not activate.

Natural-Language Policy Violations

Low

Confidence: 95% confidence
Finding: The manifest sets `"language": "en"`, which is a natural-language locale constraint. Under the policy, forcing a specific language without user opt-in or a clearly documented justification is a violation, and no such opt-in or justification appears elsewhere in this file.

Static analysis

No suspicious patterns detected.