QMT

Security checks across malware telemetry and agentic risk

Overview

This QMT trading skill is mostly documentation, but it teaches live trading and money-transfer actions with weak safety and credential-handling guidance.

Install only if you intentionally want an agent to assist with QMT brokerage workflows. Keep it in Review because connected examples can submit trades or move funds; require explicit per-action user approval, prefer paper/backtest modes, restrict accessible accounts, and never hardcode or log bank passwords or real account identifiers.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (8)

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The documented capability set exceeds the stated skill description by including bank/broker fund transfers and bank balance queries, which are materially more sensitive than ordinary trading-terminal functions. This can mislead users, reviewers, or policy gates into granting a higher-trust integration than intended, increasing the chance of unauthorized money movement or exposure of financial account data.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The document includes concrete buy/sell order examples that appear directly usable in a live trading environment, but it does not warn readers that these APIs may execute real-money trades if run against a funded account. In a trading skill, omission of safety context materially increases the chance of accidental financial loss from copy-pasted example code or misuse by inexperienced users.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The README includes a concrete live-trading example that places an order via `order_shares()` but does not clearly warn that the snippet can execute real-money trades in a broker-connected environment. In a trading skill, this omission materially increases the chance that a user copies the example into production or a live QMT session and triggers unintended market orders.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill explicitly documents live order placement functions such as buying, selling, target-value orders, and cancellation, but does not place an immediate, prominent warning near those examples that execution may affect a real brokerage account. In the context of a trading skill tied to broker-enabled QMT, an agent or user could copy or invoke these snippets against a live account, causing unintended financial loss through accidental trades rather than traditional code execution abuse.

Missing User Warnings

Medium

Confidence: 86% confidence
Finding: The document presents live trading, cancellation, and fund-transfer operations as routine API calls without explicit warnings, confirmation expectations, or discussion of irreversible financial consequences. In a high-impact financial context, this increases the risk that an agent or developer will integrate or invoke dangerous operations without appropriate safeguards.

Missing User Warnings

High

Confidence: 97% confidence
Finding: The examples normalize entering bank account numbers and plaintext bank/fund passwords directly into code and logs, which encourages insecure credential handling in a financial environment. If copied into production or repositories, this can expose credentials, enable unauthorized transfers, and compromise real brokerage/bank accounts.

Unpinned Dependencies

Low

Category: Supply Chain
Content: xtquant>=1.0.0 numpy>=1.20.0
Confidence: 94% confidence
Finding: xtquant>=1.0.0

Unpinned Dependencies

Low

Category: Supply Chain
Content: xtquant>=1.0.0 numpy>=1.20.0
Confidence: 97% confidence
Finding: numpy>=1.20.0

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal