ClawHub
Back to skill

Security audit

首板隔夜策略

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed simulated trading-analysis workflow that writes local reports and only performs simulated orders under explicit user conditions.

Install only if you are comfortable with the skill creating local strategy reports and optional SVG charts in your workspace, and review any market, plan-review, or sim-trade integrations it calls because those provide the actual data and simulated execution authority.

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 Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding
The skill clearly instructs reading multiple local reference files and writing a report into the user's workspace, yet it declares no corresponding permissions. This creates a transparency and consent problem: the runtime may still permit these actions, but users and policy layers cannot accurately assess or constrain the skill's filesystem access.

Tp4

High
Category
MCP Tool Poisoning
Confidence
83% confidence
Finding
The manifest presents the skill as a decision/evaluation tool, but the documented behavior extends to technical-indicator computation, chart rendering, and file-generation side effects. That mismatch can mislead users and security reviewers about what code paths and local capabilities are exercised, increasing the chance of overbroad trust and insufficient sandboxing.

Description-Behavior Mismatch

Low
Confidence
90% confidence
Finding
A skill framed as analysis-only should not silently create files in the user's workspace without prominent disclosure and consent. Even if the target path is limited to a report directory, automatic writes can overwrite user expectations, leak sensitive analysis artifacts locally, or be abused as a stepping stone for unwanted persistence.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
Mandating automatic report creation without a user-facing warning or consent is a real security and privacy concern. It causes a side effect outside the conversational response, may persist potentially sensitive strategy content on disk, and can surprise users who expected analysis-only behavior.

Static analysis

No suspicious patterns detected.