ClawHub
Back to skill

Security audit

Low Token Usage

Security checks for vulnerabilities and agentic risk

Overview

This skill is not malware, but it can make the agent skip confirmations and reduce visibility before taking actions.

Review this skill before installing if you rely on confirmations or detailed summaries before changes. It is best suited only where the agent must conserve tokens and where separate safety rules still require explicit approval for risky, destructive, or externally visible actions.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
88% confidence
Finding

"Skip confirmation messages" encourages the agent to make execution decisions without obtaining user confirmation at critical points. Even if intended for token conservation, this shifts the skill toward autonomous action in situations where user validation should remain mandatory.

Content

Scanner excerpt · SKILL.md (reported line 19)May include surrounding context.

md
- No preamble/postamble
- Combine multiple tool calls into one
- Prefer file edits over reads when content is known
- Skip confirmation messages

### EXTREME (When tokens < 25% remaining)
- Absolute minimum words

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The instruction to skip confirmation messages reduces user visibility into actions the agent is about to take, which can lead to unreviewed or unintended operations. In an agent skill context, this weakens an important safety and consent checkpoint, especially when combined with file edits or other side-effecting actions.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The extreme mode explicitly favors action-only behavior, minimal explanation, batching changes, and using exec over more transparent file operations, which materially reduces auditability and user awareness. This creates a higher risk that impactful operations occur with insufficient scrutiny, making mistakes or abuse harder to detect before damage is done.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.