ClawHub
Back to skill

Security audit

Humanize AI

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed local text analysis and cleanup skill, with the main practical risk being accidental overwrites during batch rewriting.

Use report-only analysis first and review rewritten output before replacing originals. Be especially careful with the documented batch command because it overwrites matching .txt files in place and does not create backups.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The code in scripts/analyze.py is an analyzer/reporting utility. It loads pattern lists, scans text for AI vocabulary, puffery, chatbot artifacts, hedging, em dashes, curly quotes, and replaceable phrases, then prints a human-readable or JSON report. While it identifies phrases that could be replaced, it never performs replacements or writes modified output. There is also no batch-processing logic, no multi-file traversal, and nothing specifically related to bypassing AI detectors beyond counting heuristic patterns. The detection/reporting portion aligns with the description, but the core declared behavior of automatically fixing/humanizing text is not present, making this a material description-behavior mismatch.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The description uses broad activation language such as using the skill whenever a user wants to analyze text, batch-process files, or automate cleanup, which can cause the agent to invoke Shell-capable file-modifying behavior in situations where the user's intent is ambiguous. In this context, the danger is increased because the skill advertises automated text rewriting and includes Shell among allowed tools, creating a realistic path to unnecessary or overbroad file operations.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The batch-processing instructions show bulk in-place modification of user files using temporary outputs and mv, but provide no warning about backups, dry runs, or manual review. This is dangerous because it can lead to irreversible corruption or unintended semantic changes across many files, especially given the skill's purpose of aggressive automated text normalization.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.