ClawHub
Back to skill

Security audit

企业报表合并·模板填充·分析报告生成

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed local financial-report merging skill with no evidence of hidden access, exfiltration, persistence, or destructive behavior.

Install only if you intend to process enterprise or financial reports. Provide only the specific files needed, review generated Excel/Word outputs before relying on them, and manually verify any values extracted from scanned PDFs or screenshots.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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 (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding
The skill clearly instructs the agent to read user-provided local files such as Excel, PDF, and Word documents and to process them via scripts, yet it declares no explicit permissions. This creates a permission-model mismatch: an agent may access sensitive enterprise financial documents without transparent declaration or user-governed scoping, increasing the risk of overbroad file access and unsafe handling of confidential data.

Intent-Code Divergence

Medium
Confidence
90% confidence
Finding
The guide first warns that image-based PDF recognition is unreliable and should not be auto-filled into Word templates, but later states screenshot recognition can achieve 'near 100%' or 'close to 100%' accuracy and says item-by-item checking is no longer needed. In a financial reporting skill, this contradiction can cause operators to over-trust OCR-derived numbers and produce materially incorrect merged reports or analysis outputs.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The trigger phrases are very broad and overlap with common enterprise document-handling requests such as merging reports or filling templates. In an agent environment, this can cause the skill to activate in situations beyond the user's intended scope, leading to unnecessary access to sensitive financial documents or unintended report generation workflows.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The trigger phrases are generic, conversational requests such as merging reports or filling a Word template, which can plausibly appear in ordinary business chat and unintentionally activate the skill. Because the skill can ingest multiple files and generate consolidated reports automatically, accidental activation could lead to unintended processing of sensitive financial documents or unexpected report generation.

Static analysis

No suspicious patterns detected.