Security audit

多邮箱管理专业版

Security checks for vulnerabilities and agentic risk

Overview

This email-management skill is broadly aligned with its stated purpose, but it asks for high-impact mailbox powers without enough enforceable safeguards.

Review this skill carefully before installing. Use it only with mailbox accounts and scopes you intend to manage, keep dry-run and previews mandatory for bulk operations, restrict callback URLs, and verify how the porteden CLI stores credentials, audit logs, and exported email data.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Intent-Code Divergence

Medium

Confidence: 90% confidence
Finding: The skill advertises exec capability, external callback URLs, and command-driven behavior, while its safety section only generically claims that only whitelisted commands are executed and user input is not concatenated into commands. Because no actual whitelist, argument validation rules, or callback restrictions are defined, an agent may over-trust the documentation and execute unsafe command compositions or reach attacker-controlled endpoints.

Intent-Code Divergence

Medium

Confidence: 85% confidence
Finding: The document claims outputs will not contain secrets, but the same skill promotes audit logs, exports, multi-account operations, and message handling across enterprise mailboxes without specifying masking or redaction controls. In this context, logs and exports can easily include email contents, addresses, message IDs, tokens in headers, or other sensitive metadata, so the assurance is misleading and unsafe.

Missing User Warnings

High

Confidence: 94% confidence
Finding: The skill documents destructive operations such as batch delete, archive, forward, reply, and marking actions in an enterprise mailbox context without strong risk warnings, confirmation requirements, or default dry-run safeguards. In a multi-account email management tool, accidental or prompt-induced execution could cause irreversible data loss, mass messaging, compliance violations, or cross-account mistakes at significant scale.

Static analysis

No suspicious patterns detected.