The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides Python and shell patterns for executing Planemo commands and ffmpeg recordings. In planemo-reference.md, it uses os.system() to launch workflows, correctly advising that variables should be quoted to prevent shell injection, although this method is less secure than structured subprocess execution.
[DYNAMIC_EXECUTION]: Includes a utility script in automation-patterns.md that searches the user's home directory (Path.home()) for Playwright's Chromium installation. This is a functional requirement to enable browser-use automation via Chrome DevTools Protocol (CDP) across different operating systems.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external Galaxy API responses and user-provided Job YAML files. This creates a standard surface for indirect injection, though the author mitigates this by recommending yaml.safe_load() and implementing status polling with terminal state validation.
[CREDENTIALS_UNSAFE]: Discusses the management of Galaxy API keys, specifically identifying the risk of using administrative keys ($MAINKEY) which provide access to all users' data. The skill promotes safe practices by instructing the user to store keys in environment variables and provides a mask_api_key function to prevent accidental exposure in log files.

galaxy-automation