Security audit

Skill开发

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent skill-development helper, but users should be aware it may create local files or code-hosting PRs as part of its normal workflow.

Install only if you want an agent to help draft and scaffold skills. When using it, specify the exact workspace path and repository target, and review generated files or PRs before accepting them.

SkillSpector

Vulnerability Patterns

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
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Tp2

High

Category: MCP Tool Poisoning
Confidence: 85% confidence
Finding: Mixing characters from multiple Unicode scripts in a single identifier is a common technique to create visually ambiguous tool names.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The template explicitly instructs skills to write output to a local file by default when connectors are unavailable, but provides no requirement to notify the user, obtain consent, or constrain the file path. In an agent-skill context, this can lead downstream skill authors to create filesystem-writing behavior that surprises users, overwrites local data, or leaves sensitive artifacts on disk.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: This second template repeats the same unsafe design pattern by making local file output the default fallback without warning or permission language. Because this is a reference template intended for reuse, the issue can propagate across many generated skills and normalize silent filesystem side effects.

Static analysis

No suspicious patterns detected.