ClawHub
Back to skill

Security audit

Multi Agent Coordinator Zhuyu28

Security checks for vulnerabilities and agentic risk

Overview

This skill is a small multi-agent coordination helper whose local file persistence is purpose-aligned, but users should avoid putting secrets into saved session data.

Install only if you are comfortable with coordination tasks and results being saved locally when the helper script is used. Do not include secrets, credentials, private prompts, or confidential plans in task descriptions or results unless you control the session file location and cleanup.

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
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Lp3

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding
The skill metadata does not declare any explicit tool scope or permissions, yet the skill was detected as having capabilities related to file access and MCP. In a multi-agent orchestration skill, undeclared file_read/file_write and external coordination capabilities increase the risk of over-privileged execution, unintended data access, and unsafe agent-to-agent or external system interactions because consumers cannot accurately assess what the skill may do.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The coordinator persists full session state to a local JSON file, including master task text, subtask descriptions, agent identifiers, and task results, without any access controls, minimization, or notice that potentially sensitive workflow content will be stored on disk. In a multi-agent orchestration context, these fields can easily contain secrets, internal prompts, proprietary plans, or user data, so silent persistence increases the risk of unintended disclosure from shared systems, backups, or later file reuse.

Static analysis

No suspicious patterns detected.