Back to skill

Security audit

writing-skills

Security checks for vulnerabilities and agentic risk

Overview

This skill is not clearly malicious, but it asks for broad command and API authority that is poorly scoped for a writing/skill-development aid.

Install only if you are comfortable with a broadly scoped development automation skill that may read files, run commands, and use API credentials. Before using it, require explicit confirmation for command execution or API calls, avoid giving it broad credentials, and limit it to a sandboxed project directory.

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)

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The manifest presents the skill as a writing/skill-creation aid, but the body broadens behavior into generic automation, API use, file operations, and command execution. This scope mismatch can cause the agent to invoke higher-risk capabilities than a user would reasonably expect, increasing the chance of unsafe actions or privilege misuse.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

Declaring exec capability for a writing/development skill is dangerous because it enables shell or process execution without a clear, necessary link to the advertised function. If the skill is triggered on ordinary writing or editing tasks, this can expose the host to command injection, destructive commands, or unintended local system access.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The skill documents external API connectivity and credential setup without clearly tying that network access to its stated purpose. Unnecessary network capability expands the attack surface by enabling data exfiltration, contact with untrusted endpoints, or misuse of stored API keys.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger condition 'when the user needs to perform related operations' is so broad that it can cause over-activation in vague or unrelated contexts. In combination with read and exec capabilities, ambiguous triggering increases the risk that the skill runs privileged actions without sufficiently specific user intent or review.

Content

No source excerpt is available for this finding.