The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection due to its processing of external, uncontrolled data. \n- Ingestion points: Untrusted data enters the context from news APIs (Tavily, SerpAPI, Brave) and market data providers (AkShare, Tushare, YFinance) as described in SKILL.md. \n- Boundary markers: The provided documentation does not indicate the use of specific delimiters or system instructions to ignore embedded commands within the fetched news content. \n- Capability inventory: The skill is capable of performing network operations (sending reports to notification channels), managing local files (configuring .env), and running a local web server for the dashboard. \n- Sanitization: No evidence of input sanitization or filtering for the external news data is present in the skill's operational overview. \n- [EXTERNAL_DOWNLOADS]: The skill relies on external code and package repositories. \n- Instructions require cloning a repository from https://github.com/ZhuLinsen/daily_stock_analysis. \n- The installation process involves downloading and installing third-party Python packages like akshare, yfinance, and litellm from public registries. \n- [COMMAND_EXECUTION]: The skill requires manual execution of shell commands and scripts for its primary functionality. \n- Setup and operation involve commands such as git clone, pip install, docker run, and executing Python scripts like main.py or web_app.py to launch the analysis engine and dashboard.

daily-stock-analysis