Security audit

技能工具

Security checks for vulnerabilities and agentic risk

Overview

This Hugo publishing skill is understandable, but it can read local memory/config files and push blog changes to git without clear approval checkpoints.

Install only if you want an agent to manage a local Hugo blog repository. Before use, require the agent to show the exact files and git commands first, and give separate explicit approval before any commit or push. Avoid letting it read broad memory files unless you know they do not contain secrets or private notes.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Description-Behavior Mismatch

Medium

Confidence: 93% confidence
Finding: The manifest description is generic and omits the skill's real capabilities: reading local config/memory files, creating or modifying Hugo content, and performing git commit/push actions. This can mislead users or orchestration systems into invoking the skill without understanding that it changes repository state and may publish content remotely.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The skill description is broad and underspecified, which increases the chance of accidental activation in contexts where the user did not intend repository modification or publication. For a skill with exec and git push capabilities, vague activation criteria materially raise the risk of unintended destructive or externally visible actions.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The quick-start tells the agent to invoke the skill with 'necessary input parameters' but does not define safety gates, approval boundaries, or when publishing is appropriate. Because the skill includes repository writes and git push, this ambiguity can lead to premature execution and unintended publication.

Missing User Warnings

High

Confidence: 97% confidence
Finding: The skill describes automatic file creation and git push without an explicit confirmation checkpoint. In context, this means an agent could write content, commit it, and publish it to a remote repository or deployment pipeline, causing unauthorized changes, accidental disclosure, or public release of unreviewed material.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The workflow explicitly reads MEMORY.md/USER.md and writes repository files, but it lacks a clear privacy notice and modification warning. This is dangerous because user memory files can contain sensitive paths, credentials, or personal data, and the skill could propagate derived or sensitive information into posts or commits without the user's awareness.

Static analysis

No suspicious patterns detected.