The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted source code provided by users to calculate metrics and suggest readability improvements.
Ingestion points: The agent reads user-specified files (e.g., in SKILL.md workflow steps) to process them through various analysis tools.
Boundary markers: There are no explicit instructions for the agent to use boundary markers or to ignore potential instructions embedded within the user's code.
Capability inventory: The skill utilizes file system read/write operations and command execution to perform its primary function of code restructuring.
Sanitization: The provided scripts analyze the code via the ast module, but the overall agent workflow relies on the content of the untrusted code to determine its actions.
[EXTERNAL_DOWNLOADS]: The skill performs complexity analysis using uvx lizard, which downloads the lizard package from the Python Package Index (PyPI) at runtime.
[DYNAMIC_EXECUTION]: Dynamic execution patterns are found within the skill's example and test files.
Evidence: assets/examples/zen-of-python/python/test_examples.py uses exec() in a test case to demonstrate the impact of star imports on the local namespace.
Evidence: assets/examples/zen-of-python/python/zen_before.py uses __import__() as part of a deliberate anti-pattern used to verify the skill's Zen of Python scanner.

write-readable-code