ClawHub
Back to skill

Security audit

博客发布工具

Security checks for vulnerabilities and agentic risk

Overview

This blog publishing skill does what it claims, but it gives the agent broad write, command, and push authority without enough user-controlled checkpoints.

Review this skill before installing. Use it only for a blog repository you intend to modify, and require the agent to show the exact blog path, files, commit message, and git remote before any write, commit, or push. Avoid letting it read memory files unless you are comfortable with local profile or configuration details being used for blog setup discovery.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (3)

Vague Triggers

High
Confidence
94% confidence
Finding
The activation guidance is extremely broad ('提升效率、自动化流程、批量处理、工作流优化') and overlaps with generic productivity requests, making accidental invocation likely. In context, this is more dangerous because the skill has read/write/exec capabilities and includes repository-modifying and git-push behavior, so a user could trigger high-impact actions without clearly intending to publish or push blog content.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The skill documents automatic file creation, git add/commit, and git push without a clear consent checkpoint or warning that repository contents and remotes will be modified. This is dangerous because it can cause unintended publication, data loss, or leaking sensitive/internal content to a remote repository, especially given the skill's exec capability and its default automation-oriented framing.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill instructs the agent to read MEMORY.md/USER.md and .git metadata to auto-discover blog configuration without warning the user that personal memory files and repository metadata may be inspected. This creates a privacy and data-minimization issue, and in context it is riskier because those sources can contain secrets, internal paths, usernames, remotes, or other sensitive environment details unrelated to the immediate task.

Static analysis

No suspicious patterns detected.