The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (CRITICAL): A confirmed detection of an untrusted remote code execution pattern was found. The skill executes a script from https://astral.sh/uv/install.sh by piping the output of curl directly into the shell (| sh). This is a high-risk pattern that allows a remote server to execute arbitrary commands on the user's system.
[CREDENTIALS_UNSAFE] (HIGH): The documentation explicitly instructs users to set sensitive credentials—including ESUN_ACCOUNT_PASSWORD, SHIOAJI_CERT_PASSWORD, and MASTERLINK_ACCOUNT_PASS—directly within Python code using os.environ. This practice encourages users to store plain-text secrets in source files, which are often accidentally committed to version control or shared.
[EXTERNAL_DOWNLOADS] (MEDIUM): The skill depends on third-party libraries esun-trade and shioaji which are not from the pre-approved trusted organizations list. These packages are installed via pip and interact with financial accounts, posing a supply-chain risk.
[DATA_EXPOSURE] (MEDIUM): The skill handles sensitive file paths for trading certificates (e.g., ESUN_CONFIG_PATH, SHIOAJI_CERT_PATH, MASTERLINK_CERT_PATH). While no explicit exfiltration was detected in the reference file, the management of these paths in scripts increases the surface area for credential theft.

finlab