Security audit

财务

Security checks for vulnerabilities and agentic risk

Overview

This finance skill should go to Review because it handles sensitive accounting and tax workflows with command/write authority while giving broad scope and unsupported security assurances.

Install only if you are comfortable giving an agent command and file-write authority around sensitive financial data. Treat generated tax forms, invoices, audit reports, and accounting entries as drafts requiring qualified human review, and do not assume the listed encryption, access-control, logging, or compliance claims are enforced by this skill.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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

Findings (5)

Context-Inappropriate Capability

Medium

Confidence: 92% confidence
Finding: The skill exposes `exec` alongside broad documentation that frames the package as a general-purpose automation aid rather than a tightly scoped finance-only workflow. In an agent setting, generic execution capability increases the chance that user-provided financial data, file paths, or generated commands could be turned into unintended local command execution or over-broad actions.

Context-Inappropriate Capability

Medium

Confidence: 85% confidence
Finding: The documentation introduces API-key handling and network/API-style behavior without clearly tying it to a specific audited finance service or limiting outbound interactions. In combination with write/exec capabilities, this can normalize workflows that send sensitive accounting data to external endpoints or expose secrets through command-line/environment usage.

Intent-Code Divergence

Medium

Confidence: 89% confidence
Finding: The file claims encryption, access control, audit logging, and compliance properties, but the markdown provides no implementation or enforceable controls, while the skill also exposes read/write/exec tooling. These unsupported assurances may cause users to trust the skill with highly sensitive financial and tax data under false security assumptions.

Vague Triggers

Medium

Confidence: 80% confidence
Finding: The invocation language is broad and ambiguous, encouraging use for general data analysis, visualization, and workflow tasks beyond a constrained finance/accounting scope. In an agent ecosystem, vague triggers can cause the skill to be selected in inappropriate contexts where its exec-enabled behavior and sensitive-data handling become riskier.

Missing User Warnings

Medium

Confidence: 87% confidence
Finding: The skill describes processing bank statements, taxes, invoices, and financial reports—data classes that are highly sensitive and can have legal and financial consequences—without prominent up-front warnings about confidentiality, verification requirements, or the risk of incorrect generated documents. This is especially dangerous because users may rely on automated outputs for tax or invoice actions without human review.

Static analysis

No suspicious patterns detected.