Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

TradingAgents-CN 股票分析助手

v1.2.1

多智能体大语言模型金融交易分析助手。基于 TradingAgents-CN 框架,使用多个专业AI分析师协作分析股票(A股/港股/美股),生成投资建议和专业报告。 触发场景: - 用户说"分析某只股票"、"帮我看看茅台"、"股票怎么样" - 用户说"股票分析"、"投资建议"、"股票研究报告" - 用户提到股票代码...

0· 104·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
Name/description (multi‑agent stock analysis) matches the included scripts and documentation. However, the code hardcodes a Windows path (E:/TradingAgents-CN) and the SKILL.md uses a generic {PROJECT_DIR} placeholder — this mismatch is surprising and reduces portability. The skill's manifest declares no required env vars but the project clearly expects multiple LLM and data API keys.
Instruction Scope
SKILL.md instructs the agent/user to clone the upstream TradingAgents-CN repo, create a .env with API keys, and run python scripts — all expected for this purpose. A minor concern: the doc tells the AI to 'remember' the project path (persistent memory) and the scripts will load .env and call external LLM/data providers, which means user inputs and tickers will be sent to third‑party APIs. The instructions do not ask the agent to read unrelated system files, but they do rely on environment variables and local repo contents that are not declared in the skill metadata.
Install Mechanism
No install spec is present (instruction-only), so nothing is fetched automatically by the skill. The code recommends running 'pip install -r requirements.txt' which is normal for a Python project; the installer risk is left to the user and not automatic.
!
Credentials
The skill metadata lists no required environment variables, but SKILL.md, references/api-keys.md, check_env.py and analyze.py expect multiple API keys (DEEPSEEK_API_KEY, DASHSCOPE_API_KEY, OPENAI_API_KEY, GOOGLE_API_KEY, TUSHARE_TOKEN, FINNHUB_API_KEY, etc.) and even optional DB creds in examples. This mismatch (declared none vs required many) is a notable incoherence and increases the risk that secrets will be provided without the user realizing the skill needs them.
Persistence & Privilege
always:false and normal autonomous invocation settings. The only persistence-related text is a suggestion that the AI 'remember' the project path; the skill does not request system-wide config modifications or cross-skill privileges. No 'always:true' or other elevated flags are present.
What to consider before installing
Things to check before installing or running this skill: - Be aware the included scripts expect you to clone the upstream TradingAgents-CN repo into a local folder; the code hardcodes E:\/TradingAgents-CN which will fail on non‑Windows systems or if you choose a different path — inspect and adjust paths before running. - The skill metadata declares no required env vars, but the docs and scripts require multiple API keys (LLM providers and market data APIs). Only provide the minimum API keys you intend to use, and never commit your .env to git. - Running the scripts will send data (stock tickers, prompts, possibly snippets of reports) to third‑party LLM and data provider endpoints — do not include any private or sensitive information in prompts or configuration. - Because the skill is instruction-only and will instruct you to 'pip install -r requirements.txt', review requirements.txt and the cloned repository code before installing dependencies; run in an isolated environment (virtualenv/container) if possible. - The mismatches (no declared env vars, hardcoded path, vague 'AI will remember' note) suggest sloppy packaging rather than overt malice, but exercise caution: review the upstream TradingAgents-CN repository and the included scripts thoroughly before use. If you want, I can summarize the exact lines you should change (path and env handling) to make this safer and portable, or produce a checklist to run the scripts in a sandboxed environment.

Like a lobster shell, security has layers — review code before you run it.

latestvk976t4sq5gfcp8gs1gk19qnsyh83yktt

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Comments