The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines commands to execute test suites using pytest and coverage (e.g., pytest tests/unit/ -v). This is standard functionality for a testing skill and operates within the expected development environment.
[DATA_EXPOSURE]: The skill references sensitive files like .env and tests/cred.py. It explicitly advises against committing secrets to version control and recommends using environment variables, which is a recognized security best practice for secret management.
[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it executes code within the tests/ directory.
Ingestion points: The agent reads and executes files from the local tests/ directory.
Boundary markers: None present for the code being executed.
Capability inventory: Shell command execution via pytest and coverage (SKILL.md).
Sanitization: No sanitization is performed on the test scripts before execution.
Analysis: Since this risk is inherent to the primary purpose of the skill (testing code) and the skill includes hygiene warnings, the risk is considered acceptable for its intended use case.

testing