Back to skill

Security audit

Us Stock Radar Gold3bear

Security checks for vulnerabilities and agentic risk

Overview

This skill fetches public U.S. market data and provides manual news/social lookup links, with some documentation gaps but no hidden persistence, credential access, or destructive behavior.

Before installing, expect the skill to run Python scripts that contact public finance and news/social sites and to produce Chinese-language output. Treat its X/Twitter and Google News portions as manual lookup aids, not a complete sentiment-analysis system, and only schedule recurring checks if you are comfortable with repeated outbound market-data requests.

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
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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
Findings (18)

Tp4

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

The declared description emphasizes both market quote retrieval and news/social sentiment monitoring for U.S. stocks. However, this code chunk only orchestrates a dashboard by invoking local scripts for index/futures/commodity data and big-tech stock quotes. There is no visible implementation of Google News RSS or X/Twitter monitoring here. Additionally, the dashboard includes a specific tech-giant quotes feature that is not clearly called out in the description, which focuses more on indexes, futures, and general U.S. market queries. While this chunk may be only part of a larger skill, based on the supplied code alone the behavior does not fully match the declared capabilities.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The description emphasizes a broad 'US stock market and sentiment monitoring' skill triggered by questions about Nasdaq, S&P, Dow, US market conditions, and futures, with explicit support for Yahoo Finance plus Google News/X sentiment sources. The actual code chunk only defines a fixed list of eight tech stocks and passes them to us_quote.py. That is materially narrower and different in purpose from monitoring the overall US market. Also, the advertised sentiment/news monitoring capability is not present in this code chunk. While this script may be a component of a larger finance tool, based on the supplied code it does not accurately represent the declared behavior.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The description says this skill is for US stock market行情与舆情监控, specifically claiming Yahoo Finance quotes plus Google News RSS and X/Twitter sentiment monitoring. The supplied code only implements market data retrieval and console output. It fetches index/futures/commodity quotes from Yahoo Finance, falls back to Finviz scraping, and also retrieves NY Fed overnight rates and FRED Treasury yields. There is no code for Google News RSS, no Twitter/X access, and no sentiment analysis. Therefore the declared description does not accurately represent the actual behavior. The extra macro/commodity sources are materially different additions beyond the declared purpose, while the missing sentiment functionality is the clearest mismatch.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

该代码的实际功能较窄:它调用 Yahoo Finance 的 chart 接口查询指定股票代码的当日价格与涨跌信息,并输出结果。虽然这与声明中的“支持Yahoo Finance获取实时行情”部分一致,但声明的核心范围更广,包含美股大盘/指数/期货查询场景,以及 Google News RSS 和 X/Twitter 舆情监控;这些能力在提供的代码中都没有体现。因此描述未准确代表该代码块的实际行为,属于明显不一致。

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
84% confidence
Finding

The skill documents executable Python and shell usage plus outbound network access, but it does not declare any explicit tool scope such as allowed-tools or permissions. In an agent environment, missing scope boundaries can cause the skill to run with broader capabilities than users or the platform expect, increasing the blast radius if the skill is modified, misrouted, or invoked in the wrong context.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

Overly broad trigger phrases can cause accidental invocation during ordinary market conversation, making the agent fetch network data or follow finance-analysis workflows when the user did not intend to use this skill. In multi-skill environments, ambiguous routing increases the chance of unnecessary external calls, privacy leakage in prompts, and execution of capabilities outside the narrowest-necessary context.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The manifest description and title are entirely in Chinese and present the skill as operating in Chinese, with no indication that the user may choose another language. Under the policy, a locale or language constraint should not be forced without opt-in unless clearly documented and justified.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/dashboard.py (reported line 16)May include surrounding context.

python
print("=" * 55)

print("\n>>> 主要指数 + 期货 + 大宗商品")
subprocess.run([sys.executable, os.path.join(base, "us_index.py")])

print("\n>>> 科技巨头 (NVDA/AAPL/MSFT/GOOGL/AMZN/META/TSLA/AMD)")
subprocess.run([sys.executable, os.path.join(base, "us_quote.py"),

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/dashboard.py (reported line 19)May include surrounding context.

python
subprocess.run([sys.executable, os.path.join(base, "us_index.py")])

print("\n>>> 科技巨头 (NVDA/AAPL/MSFT/GOOGL/AMZN/META/TSLA/AMD)")
subprocess.run([sys.executable, os.path.join(base, "us_quote.py"),
                "NVDA", "AAPL", "MSFT", "GOOGL", "AMZN", "META", "TSLA", "AMD"])

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/tech_giants.py (reported line 14)May include surrounding context.

python
if __name__ == "__main__":
    tech_stocks = ["NVDA", "AAPL", "MSFT", "GOOGL", "AMZN", "META", "TSLA", "AMD"]
    print("=== 苹果+微软+谷歌+Meta+英伟达+特斯拉+AMD ===")
    subprocess.run([sys.executable, os.path.join(base, "us_quote.py")] + tech_stocks)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The module description states the skill provides market data in Chinese, and the rest of the user-facing labels and console messages are exclusively Chinese. This creates a language policy concern because the skill appears to enforce a specific language without offering the user a locale or language choice.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The manifest describes a US stock market monitoring tool focused on Yahoo Finance quotes and Google News/X(Twitter) sentiment. This file additionally pulls NY Fed reference rates and FRED Treasury yield data, which are macroeconomic datasets not mentioned in the manifest and broader than the stated行情与舆情监控 scope.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The module docstring is entirely in Chinese and presents usage only in Chinese, which imposes a specific language on users without any opt-in or explanation. The file also appears to be a general stock-quote utility rather than a clearly region-specific compliance tool, so the locale restriction is not obviously justified.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The manifest says the skill supports Yahoo Finance, Google News RSS, and X/Twitter sentiment monitoring. In this file, executable examples are provided for Yahoo Finance, NY Fed, and FRED, while the X/Twitter section only instructs use of a browser to visit search URLs, with no actual monitoring or analysis code shown. That is a mild description-to-behavior mismatch because the claimed capability is not actually implemented here.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

The description emphasizes US equity market quotes and sentiment monitoring, especially major indexes, futures, and related news. The file also adds NY Fed rates, FRED Treasury yields, WTI crude, gold, and silver tracking, which broaden the capability into macro and cross-asset monitoring rather than just US stock market radar. While related, this is broader than the manifest's stated purpose.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The module docstring and user-facing console labels are written entirely in Chinese, which imposes a specific language on users without any opt-in or documented locale constraint. The policy specifically calls for flagging language or locale restrictions when no user choice is offered.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

The file contains user-facing natural-language text only in Chinese, including the module description and printed output. This can violate a language/locale policy when the skill imposes a specific language without user opt-in or any documented locale justification.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The top-level documentation states there are three data sources and lists only Yahoo query1, Yahoo query2, and Finviz. However, later code also calls the New York Fed API and FRED CSV endpoint for macro rates and Treasury yields, so the documentation actively misstates the implemented sources.

Content

No source excerpt is available for this finding.