The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes pytest with coverage flags to generate reports, which is a common and safe practice in software development.\n- [INDIRECT_PROMPT_INJECTION]: The agent is instructed to review annotated source code files from the coverage report. This presents a surface for potential instructions embedded in code comments to be read by the agent. 1. Ingestion points: Annotated source files in the cov_annotate directory. 2. Boundary markers: Not specified. 3. Capability inventory: pytest command execution and file system access for creating tests. 4. Sanitization: Not mentioned.\n- [DYNAMIC_EXECUTION]: The agent performs code generation by adding tests to cover missing lines, which is the primary and legitimate purpose of the skill.

pytest-coverage