Backtest Engine - Run Backtests

• Outcome: Review
• Latest audit: May 29, 2026 at 9:27 PM UTC
• Version:
  0.1.0

Overview

This appears to be a normal trading backtest tool, but its custom strategy option runs user-supplied Python code without making that risk clear.

Install only if you are comfortable with a local Python backtesting tool. Use built-in strategies or CSV/Yahoo Finance inputs normally, run it in a virtual environment, and treat any third-party --strategy-file as executable code that should be reviewed before running.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 88% confidence
• Finding: The skill explicitly describes generating JSON and Markdown reports, which implies filesystem write capability, yet `tools_available` is empty and no permission model is declared. This mismatch can cause the agent to perform file-writing side effects without clear authorization boundaries or user awareness, increasing the risk of unintended artifact creation or overwriting files in the working environment.

Context-Inappropriate Capability

High

• Confidence: 99% confidence
• Finding: load_custom_strategy() imports and executes arbitrary Python from a user-provided file path using exec_module(), which grants full code execution with the privileges of the process. In a backtesting skill, this materially expands scope from strategy evaluation to arbitrary local code execution, enabling file access, network calls, persistence, or environment secret theft.

Missing User Warnings

Low

• Confidence: 84% confidence
• Finding: The skill states that it generates JSON and Markdown reports, including named output files, but does not disclose this as a side effect or warn the user that files may be created. In an agent environment, silent file creation can surprise users, leak sensitive strategy information into persistent artifacts, or clutter shared workspaces.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The CLI accepts an arbitrary Python file and executes it without any prominent warning that this runs external code. That omission increases the chance that users treat the input as mere strategy data rather than executable code, which can lead to accidental execution of malicious files.

VirusTotal

66/66 vendors flagged this skill as clean.