Security audit

Humanizer

Security checks for vulnerabilities and agentic risk

Overview

This is a text-editing skill for making writing sound less AI-generated, with misuse and review risks but no hidden execution, persistence, or data exfiltration behavior found.

Install only if you want an editor that rewrites text to sound less AI-generated. Review all edits before applying them, especially in academic, legal, professional, or disclosure-sensitive settings where changing voice or masking AI assistance could violate rules or expectations.

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
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill is explicitly marketed to 'remove signs of AI-generated writing,' which can enable concealment of authorship and support deception in academic, professional, or trust-sensitive contexts. The absence of any warning, policy boundary, or misuse restriction increases the likelihood that users will apply it to evade disclosure requirements or authenticity checks.

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The phrase "Or invoke directly when editing documents" does not specify what exact trigger phrases, contexts, or constraints cause the skill to activate. Because it lacks clear scope or exclusion conditions, it could overlap with normal editing activity and lead to unintended invocation.

Natural-Language Policy Violations

Medium

Confidence: 88% confidence
Finding: The skill explicitly encourages adding first-person perspective, opinions, humor, and 'personality' without conditioning that behavior on the user's intent, role, or authorship. In practice, this can cause integrity and impersonation problems by making edited text falsely appear to reflect the original author's personal views or identity, especially in professional, academic, or organizational contexts.

Missing User Warnings

Low

Confidence: 86% confidence
Finding: This markdown skill advertises `Write` and `Edit` capabilities, which can affect user data, but the description and instructions do not explicitly warn users that the skill may change file contents. For a markdown skill, user-facing disclosure about data-affecting behavior is expected when the skill can rewrite or overwrite text.

Static analysis

No suspicious patterns detected.