ClawHub
Back to skill

Security audit

less-token

Security checks for vulnerabilities and agentic risk

Overview

This is a text-only prompt-compression skill with a small wording mismatch, and I found no code, persistence, or data-access behavior.

Installers should treat this as a convenience prompt-compression aid. Be aware that one instruction may encourage direct summarization in addition to compression, so avoid giving confidential content unless you are comfortable with the host AI model processing it.

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
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Intent-Code Divergence

Medium
Confidence
97% confidence
Finding
The file states that the skill 'only translates verbose summarization prompts,' but later examples and the final response advertise performing summarization and related transformations. This inconsistency is dangerous because policy enforcement or tool selection may trust the narrower description, while the actual prompt steers the model into broader content-processing behavior.

Description-Behavior Mismatch

Medium
Confidence
94% confidence
Finding
The prompt claims to be a narrow text-to-text compressor, but the final response expands behavior to directly summarize arbitrary content using 'minimal tokens.' This creates a scope mismatch that can bypass higher-level controls, routing, or user expectations that rely on the skill being limited to prompt translation rather than task execution.

Static analysis

No suspicious patterns detected.