ClawHub
Back to skill

Security audit

Decision Distiller

Security checks for vulnerabilities and agentic risk

Overview

This skill is a local decision-record helper with disclosed markdown templates and simple local scripts, not a hidden or high-impact agent behavior change.

Install this if you want local decision records and templates. Be aware that decisions you document may persist as markdown files, so avoid recording sensitive personal or business information unless you intend to keep it locally.

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)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding
The declared purpose describes a decision-analysis/documentation skill for helping users capture and analyze decisions. The supplied code does not implement decision distillation or decision-record creation. Instead, it is a development/test utility that inspects local files and directories to verify the skill's packaging and metadata. That is a materially different primary purpose and involves undeclared filesystem-validation behavior. While such a script may support the skill's development, the code chunk itself does not match the described end-user capability.

Natural-Language Policy Violations

Low
Confidence
93% confidence
Finding
Lines L023-L025 present a full usage scenario and expected output exclusively in Chinese, while the rest of the skill is in English. This introduces a language-specific behavior/example without documenting user language choice or making the locale constraint explicit.

Static analysis

No suspicious patterns detected.