The Agent Skills Directory

[COMMAND_EXECUTION]: The CoverageAnalyzer.analyze_coverage method executes shell commands via subprocess.run using a provided test_command string. This allows for arbitrary command execution if the input is not strictly validated.
[DYNAMIC_EXECUTION]: The skill dynamically generates executable Python, JavaScript, and TypeScript test files using string templates and metadata extracted from the source code. This generated code is intended for execution within the user's environment.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill accesses local source code files using open(file_path) to perform static analysis. While this is restricted to file reading, it represents a data access capability.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted source code to generate new code, creating a surface for injection attacks.
Ingestion points: Source files are read via the analyze_file method in the TestGenerator class.
Boundary markers: None; there are no instructions to delimit or ignore instructions embedded in the code being analyzed.
Capability inventory: File reading (open), static analysis (ast.parse), and command execution (subprocess.run).
Sanitization: None; the skill directly interpolates function names, arguments, and docstrings from the target code into the generated test files.

unit-testing-test-generate