ClawHub
Back to skill

Security audit

日志脱敏工具

Security checks for vulnerabilities and agentic risk

Overview

This skill is a local log-sanitization guide with expected access to sensitive log files, but users should be careful with in-place redaction, plaintext backups, and the unexplained optional callback URL.

Use this only on log paths you explicitly choose. Run preview mode before --redact, treat generated .bak files as still sensitive, and verify any external script before running it because the inspected package did not include the referenced log-sanitizer.py. Avoid using callback_url unless you have confirmed it sends only non-sensitive completion status.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The skill repeatedly claims that processing is purely local and that no data is sent externally, yet it defines a `callback_url` input for outbound notification. In a log-sanitization context, that mismatch is security-relevant because users may supply sensitive logs under the assumption that nothing leaves the machine, while callback behavior can enable metadata leakage or future content exfiltration if implemented carelessly.

Context-Inappropriate Capability

Medium
Confidence
92% confidence
Finding
An outbound callback mechanism is unjustified for a tool whose stated purpose is local log scanning and redaction. Even if intended only for completion notifications, it expands the attack surface by introducing egress capability into a workflow likely to process secrets, credentials, and regulated data.

Vague Triggers

Medium
Confidence
81% confidence
Finding
The activation scope is described in very broad terms, including many generic security, compliance, and automation scenarios without sharp boundaries. In an agent setting, overly broad triggering can cause the skill to run on unrelated files or contexts, increasing the chance of unnecessary access to sensitive logs and accidental execution of file-modifying commands.

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The skill describes redacting files in place and creating `.bak` backups, but it does not present a prominent warning about user data modification and residual sensitive data in backups. This is dangerous because users may assume the process is low-risk, while it can alter production logs, interfere with forensic integrity, and leave original secrets recoverable in backup files.

Static analysis

No suspicious patterns detected.