ClawHub
Back to skill

Security audit

Skill Creator

Security checks across malware telemetry and agentic risk

Overview

This is a coherent skill-authoring helper with expected local file creation, validation, and packaging behavior, and no evidence of hidden network, credential, or destructive activity.

Install this if you want help authoring AgentSkills. Run its scripts only on skill directories you intend to create or package, review generated SKILL.md changes before reuse, and inspect .skill archives before sharing so private files are not included.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (5)

Lp3

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding
The skill instructs the agent to invoke local scripts, edit files, and package skills, which implies file read/write and shell execution capabilities, yet it declares no permissions or operational boundaries. This creates a transparency and governance gap: a caller may trigger actions with broader side effects than expected, including filesystem changes and script execution.

Tp4

High
Category
MCP Tool Poisoning
Confidence
84% confidence
Finding
The description frames the skill primarily as creating, editing, improving, and auditing skills, but the body also directs packaging into distributable archives and implies archive/file-walking behavior. That mismatch can cause the agent or user to underestimate that the skill may produce artifacts, traverse directories, and handle symlinks, increasing the chance of unintended data inclusion or broader filesystem effects.

Vague Triggers

Medium
Confidence
77% confidence
Finding
The trigger phrases are very broad, such as 'improve this skill' or 'review the skill', making accidental activation likely in ordinary conversations about skills or documentation. Over-broad routing raises the risk that the skill is invoked in contexts where file edits, restructuring, or script execution were not intended, especially because this skill can modify skill contents.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The trigger examples include very broad phrases such as asking to 'build' or 'help me plan', which can match ordinary conversation and cause the skill to activate outside its intended scope. In a skill that creates, edits, or audits other skills, unintended invocation is risky because it can steer the agent into restructuring files or applying skill-authoring instructions when the user only wanted general advice.

Self-Modification

High
Category
Rogue Agent
Content
- Specific triggers/contexts for when to use it
- Concrete examples of user requests that should activate it

#### Write SKILL.md Body

Keep under 500 lines. Use references/ for detailed content.
Confidence
91% confidence
Finding
Write SKILL

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.