The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts loc.py, long_functions.py, and smell_check.py use subprocess.run() to execute git ls-files. This is an expected and legitimate behavior for a tool designed to audit files within a git repository.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it is designed to ingest and analyze untrusted data from external codebases.
Ingestion points: Repository files are read into memory using Path.read_text() within the scripts/ directory.
Boundary markers: The scripts do not include specific boundary markers or instructions to ignore embedded commands when presenting analysis results or code snippets to the agent.
Capability inventory: The skill can list files via subprocess.run() and analyze code structures using the Python ast module.
Sanitization: The file content processed by the scripts is not filtered or sanitized before being analyzed or presented in the output.

coding-guidelines