The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill downloads a source archive from an unverified third-party site (prdownloads.sourceforge.net) and proceeds to extract, configure, and install it using system-level commands.
Evidence: wget http://prdownloads.sourceforge.net/ta-lib/ta-lib-0.4.0-src.tar.gz, tar -xzf ta-lib-0.4.0-src.tar.gz, ./configure --prefix=/usr, make, sudo make install.
[COMMAND_EXECUTION]: The skill uses sudo to perform system-level installations and library registrations, which is a privilege escalation pattern.
Evidence: sudo apt-get update, sudo apt-get install -y build-essential wget, sudo make install.
[EXTERNAL_DOWNLOADS]: The skill directs users to download binary wheels (.whl) from an unofficial GitHub repository if standard installation fails.
Evidence: https://github.com/cgohlke/talib-build/releases.
[CREDENTIALS_UNSAFE]: The skill prompts the user for sensitive API keys (OpenAlgo, Binance) via AskUserQuestion and writes them to a .env file. While it attempts to add .env to .gitignore, handling raw secrets through prompt input and writing them to disk presents an exposure risk if the environment is not properly secured.
Evidence: Steps 6c and 6d prompt for API keys and Secret keys which are then stored in .env.

setup