Security audit
Security checks for vulnerabilities and agentic risk
This stock-analysis skill is not overtly malicious, but it should be reviewed carefully because it gives direct trading-style guidance while overstating how complete its analysis engine is.
Install only if you want a Chinese technical-analysis helper and treat its outputs as educational signals, not financial advice. Review the incomplete placeholder implementation before relying on any buy/sell, position-size, stop-loss, or stock-screening recommendation.
技能描述与文档多处声称具备23个技术指标、完整决策规则和交易信号生成能力，但 analyze_stock 依赖的 _calculate_indicator_signal 在实现中明确写着“返回模拟数据”，固定返回 HOLD、0分、0.5置信度。实际行为只是框架/样例级引擎，而不是宣称的完整分析技能，属于明显的描述与行为不匹配。
The skill defines broad scenarios and generic search keywords such as stock analysis, buy/sell points, and market review that overlap heavily with ordinary finance queries. In an agent/router environment, this can cause the skill to activate for general conversations and steer responses into high-risk investment guidance without clear user intent or scope boundaries.
This markdown file describes an AI stock-analysis skill that generates buy/sell signals, suggested positions, stop-loss/take-profit levels, and complete trading decision flows. Under the markdown-specific SQP-2 criteria, these user-impacting behaviors should be accompanied by a clear warning about financial risk and the consequences of acting on the recommendations, but the document does not provide such a user-facing warning.
文档在算法部分明确宣称综合评分为 Σ(指标分数 × 指标权重 × 市场环境系数 × 共振系数)，但 analyze_stock 中的实际实现只是 sum(s.score for s in indicator_signals)。这会导致用户对评分机制、风险解释和输出含义产生错误预期，属于文档与代码的主动矛盾。
_load_indicators 的注释写明“返回23个指标的配置”，而 _get_indicator_config 又称“这里应该从指标数据库加载”，但实际 configs 仅包含 M001、M002 两项，其他指标会得到空字典。该注释/文档与真实代码行为直接矛盾，容易让调用方误以为23个指标均已可用。
get_indicator_database() 的 docstring 和注释都写明“返回所有23个指标的完整配置”，但返回值中的 indicators 实际只是空列表占位。这不是信息缺失，而是对返回内容的直接误导，构成文档与代码意图不一致。
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
"include_indicators": True
}
response = requests.post(endpoint, json=payload)
result = response.json()
print("单股票分析结果:")
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
"include_indicators": True
}
response = requests.post(endpoint, json=payload)
result = response.json()
print("单股票分析结果:")
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
"include_indicators": True
}
response = requests.post(endpoint, json=payload)
result = response.json()
print("单股票分析结果:")
The example input '分析一下比亚迪这只股票' is a natural everyday request and does not distinguish between normal discussion and intentional invocation of this specialized skill. That ambiguity increases the chance of accidental routing, making the agent provide structured trading advice when the user may only want general information.
The stock-screening example uses a broad phrase about finding stocks that meet '妖股启动点' conditions, but it does not define activation limits, eligibility checks, or safeguards around speculative screening. In this financial context, accidental or overbroad triggering is more dangerous because it can promote highly speculative recommendations at scale rather than a single-stock analysis.
The metadata sets language: chinese, which is a natural-language policy concern when a skill enforces a specific language without opt-in or an explicit region-specific justification. The surrounding documentation does not state that users may choose another language or that the Chinese-only constraint is mandatory for compliance or locale reasons.
No suspicious patterns detected.