Back to skill

Security audit

CSV处理器 专业版

Security checks for vulnerabilities and agentic risk

Overview

This CSV skill is mostly a local data-cleaning guide, but it includes inconsistent API/network instructions and automatic network diagnostics that do not fit the stated purpose.

Review this skill before installing in restricted or production environments. It is not clearly malicious, but treat it as a Review item because its local CSV-processing purpose is mixed with unclear API/network behavior. Use it only on copied data, write outputs to new files, avoid supplying API keys unless a specific trusted tool actually requires them, and store audit logs with the same care as the source datasets.

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (5)

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill is presented as a local CSV-processing tool, but later introduces API-oriented behavior such as LLM/API dependency and HTTP/network concepts that are unrelated to core CSV cleaning. This expands the perceived trust boundary and may mislead an agent into enabling network-capable workflows or handling credentials where none should be needed.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The documented failure-handling behavior says to automatically run ping/network diagnostics and retry commands, which is not justified for a CSV-cleaning skill. This introduces unnecessary command execution and network activity, increasing attack surface and creating opportunities for misuse in restricted or sensitive environments.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The documentation contradicts itself by claiming no extra API key is needed while also listing an LLM API as a required dependency and later describing API authentication failures. Contradictory capability and credential requirements can cause agents or users to over-grant permissions, mishandle secrets, or invoke unintended remote services.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill documents row dropping, column dropping, type coercion, merging, and export/output operations that can irreversibly alter datasets, but it does not prominently warn users about possible data loss or overwriting. In an agent setting, this can lead to destructive transformations being applied to production data without adequate confirmation or backup expectations.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The audit logging feature records transformation details to local JSONL logs, and examples show field names, replacement values, and reasons that may encode sensitive source-data context. Without a privacy warning, users may unintentionally persist regulated or confidential information in logs that are easier to overlook and broadly accessible.

Content

No source excerpt is available for this finding.