ClawHub
Back to skill

Security audit

Hugo博客发布专业版

Security checks for vulnerabilities and agentic risk

Overview

This Hugo blog skill matches its stated purpose, but it can write, commit, push, and deploy site content without clear confirmation checkpoints.

Review before installing if you do not want an agent to publish or deploy your blog. Use it only on a repository you intend the agent to modify, require explicit approval before file writes, git commit, git push, or deployment, and keep GitHub, Slack, CDN, and SSH credentials scoped to the minimum needed.

SkillSpector

Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • 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
Findings (4)

Intent-Code Divergence

Medium
Confidence
90% confidence
Finding
The skill's safety guidance says not to concatenate user input into shell commands, yet earlier examples interpolate variables directly into file paths, heredocs, and git-related shell operations. In an agent setting, if any of those variables are influenced by user-controlled input, this inconsistency can lead to command or path manipulation and unsafe file writes.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill contains examples that create or overwrite content files, modify repository state, commit changes, push to remote, and trigger automated deployment, but it does not require explicit confirmation before performing those side effects. In an agent environment, this materially increases the risk of unintended publication, destructive edits, or unauthorized deployment from a casually phrased request.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Content
extended: true
      - name: 安装图片优化工具
        run: |
          sudo apt-get install -y jpegoptim optipng
          npm install -g imagemin-cli
      - name: 优化图片
        run: |
Confidence
92% confidence
Finding
The workflow uses sudo to install packages during execution, which normalizes privileged command execution in a skill intended for agent-driven automation. In environments where the agent can execute these commands, this expands the blast radius from content management into host-level package modification and increases the risk of system compromise or unintended changes.

Chaining Abuse

High
Category
Tool Misuse
Content
hugo-version: 'latest'
          extended: true
      - name: 安装图片优化工具
        run: |
          sudo apt-get install -y jpegoptim optipng
          npm install -g imagemin-cli
      - name: 优化图片
Confidence
88% confidence
Finding
The skill chains package installation, global npm install, image processing, build, deploy, and external notification in one automation flow. In an agent context, such chaining increases the chance that a single invocation causes broad side effects across the host, repository, deployment target, and third-party services without meaningful checkpoints.

Static analysis

No suspicious patterns detected.