Agent Stock Pro

• Outcome: Review
• Latest audit: Jun 5, 2026 · 3:33 PM UTC
• Version: 0.4.0

Overview

The skill is mostly coherent for stock analysis, but it automatically persists simulated portfolio data and can send generated financial reports without clear per-run user approval.

Install only if you are comfortable with the agent installing Python packages, writing financial analysis and simulated portfolio history under dist/, and potentially sending generated PDF reports through the configured message/WeChat channel. Prefer running it in a dedicated workspace, review the generated files before enabling cron-style automation, and require manual confirmation before outbound report delivery.

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 Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (11)

Context-Inappropriate Capability

Medium

• Confidence: 91% confidence
• Finding: The skill instructs the agent to install packages and even help set up a Python environment, which expands the agent’s behavior from stock analysis into system modification. That increases risk because dependency installation can alter the host environment, pull unreviewed code from package registries, and create a path for unintended or unauthorized execution beyond the user’s immediate analysis request.

Description-Behavior Mismatch

Medium

• Confidence: 92% confidence
• Finding: The skill expands from stock screening into stateful simulated trading by reading and writing portfolio state, executing buy/sell logic, and appending trade logs across runs. That persistent side effect is security-relevant because it modifies local data and can silently create or overwrite records without explicit user awareness or run-time confirmation.

Missing User Warnings

Medium

• Confidence: 84% confidence
• Finding: The documentation mentions isolated cron-job execution for stock screening and PDF sending without clearly stating approval, destination controls, or safeguards for unattended outbound delivery. In an agent skill, automated scheduled execution plus report transmission can lead to unreviewed actions, accidental data disclosure, or persistent behavior the user did not fully understand or authorize.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The skill instructs saving a detailed holdings analysis to a local file without informing the user or requiring consent. Because holdings data can include sensitive financial information, silent persistence creates privacy and data-retention risk, especially on shared systems or multi-tenant agent environments.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The skill instructs the agent to save results to a local path under dist/trade without warning the user that it will write files to disk. Undisclosed file creation can surprise users, leak sensitive trading/account analysis into persistent storage, and create privacy or compliance issues on shared systems or managed environments.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill instructs saving analysis results to local files without warning the user that files will be created under `dist/screen/`. Silent filesystem writes are risky in agent environments because they create persistent artifacts and may overwrite prior outputs or leak sensitive analysis data to shared storage.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: This workflow not only writes files but also mutates ongoing portfolio state, appends trade logs, and saves equity snapshots, all without an explicit consent step. In an agent setting, hidden persistent state can mislead later runs, expose financial preferences/history, and cause hard-to-detect integrity issues if files are stale, corrupted, or shared.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The skill directs local PDF generation and then sending the file to the current WeChat user without warning or confirmation. That combination of local artifact creation plus external transmission is security-sensitive because it can exfiltrate portfolio analysis, trading history, or other generated content to an unintended recipient.

Missing User Warnings

Low

• Confidence: 91% confidence
• Finding: The document directs automatic creation and updating of files under `dist/strategy/` without any explicit user notice or confirmation that local state will be modified. While this is only a simulation system and the target path is scoped to a project output directory, silent state changes can still surprise users, overwrite prior results, or be chained with other automation in unsafe ways.

Missing User Warnings

Low

• Confidence: 93% confidence
• Finding: The workflow embeds automated trading-simulation updates and report modification into the daily process without a user-facing warning that files and reports will be changed. This increases the risk of unintended state mutations, especially when the step is triggered as part of a larger automated reporting pipeline.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The skill instructs the agent to write analysis results directly to `dist/trade/YYYYMMDD_hhmm_股票名称.md` without any requirement for user confirmation, path validation, or disclosure that local disk state will be modified. In an agent environment, implicit file writes can create unintended persistence, overwrite prior outputs, or leak sensitive trading/account analysis into accessible storage.

VirusTotal

64/64 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.