ClawHub
Back to skill

Security audit

Security Quant Backtest

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent educational backtesting skill with notable accuracy risks in its example engine but no hidden access, persistence, exfiltration, or destructive behavior.

Reasonable to install for educational research assistance, but do not rely on the embedded backtest engine for real trading or investment decisions without fixing and testing position accounting and execution-timing assumptions. Treat generated performance results as illustrative only.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (3)

Vague Triggers

Medium
Confidence
93% confidence
Finding
The trigger section lists generic phrases such as "strategy research," "performance analysis," and "Monte Carlo," plus broad Chinese equivalents, without stating when the skill should or should not activate. In a markdown skill description, this can cause unintended invocation during general investing or analytics conversations rather than only explicit backtesting requests.

Intent-Code Divergence

Medium
Confidence
98% confidence
Finding
The engine declares a position ledger, but the buy and sell paths never update self.position, so holdings, liquidation behavior, and portfolio valuation are internally inconsistent. This means the backtest can produce materially incorrect PnL, risk, and trade statistics while appearing to function normally, which is dangerous in a trading-research skill.

Intent-Code Divergence

Medium
Confidence
97% confidence
Finding
The skill explicitly documents T-day signal generation with T+1 execution as the correct anti-lookahead method, but the engine executes on the same date using the current close price. This creates a classic lookahead/backtest-integrity flaw that can systematically inflate reported performance and mislead users into trusting non-reproducible results.

Static analysis

No suspicious patterns detected.