The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides an evaluate command that executes arbitrary Python expressions within the context of the application being debugged.
This capability allows for complete code execution within the target process.
Evidence: In scripts/debug.py, the evaluate action calls session.inspect(expression=argv[2]).
[COMMAND_EXECUTION]: The skill manages the lifecycle of external Python processes and can execute them with arbitrary file paths provided via CLI arguments.
Evidence: The start and probe actions in scripts/debug.py launch Python scripts provided as command-line arguments.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the debugpy library, a well-known debugging tool maintained by Microsoft.
Evidence: SKILL.md instructs the user to pip install debugpy.
[PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection by accepting arbitrary expressions and file paths that are executed or processed at runtime.
Ingestion points: Command-line arguments in scripts/debug.py used for the evaluate, start, and probe actions.
Boundary markers: None provided in the command interface to distinguish instructions from data.
Capability inventory: Includes arbitrary code execution (evaluate), process management (start), and variable inspection.
Sanitization: No input validation or sanitization is performed on the provided expressions before they are passed to the debugger session.

debugger-python