The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted Python source code files (e.g., from src/ and tests/ directories) to evaluate them against a quality checklist. This creates a surface where malicious code or comments within the processed files could contain instructions intended to override the agent's review logic or influence its final assessment.
Ingestion points: Python source code, docstrings, and test files being reviewed.
Boundary markers: No specific delimiters or "ignore instructions" warnings are defined for the code ingestion process.
Capability inventory: The skill suggests executing shell commands including ruff, basedpyright, and pytest (in SKILL.md).
Sanitization: No sanitization or validation of the input source code is specified before processing or command execution.
[COMMAND_EXECUTION]: The skill provides specific shell commands for the agent to run, such as ruff format ., basedpyright, and pytest. Specifically, pytest executes the project's test suite, which involves running the Python code under review. If the code being reviewed or the test files themselves are malicious, this could lead to arbitrary code execution within the environment where the review is performed.

python-code-review