The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection (Category 8) because it is designed to ingest and analyze untrusted external data (test cases).
Ingestion points: Data enters the agent context through the instructions in prompts/test-case-reviewer.md which asks the AI to process provided test cases.
Boundary markers: The skill lacks explicit delimiters or instructions to ignore potential commands embedded within the test case content being reviewed.
Capability inventory: The skill includes several Python scripts (scripts/batch_convert_templates.py, scripts/convert_formats.py, etc.) that use subprocess.call to execute local conversion tasks. These scripts do not have network access or sensitive file system capabilities.
Sanitization: There is no evidence of sanitization or validation of the input data before it is processed by the AI.
[COMMAND_EXECUTION]: Several utility scripts in the scripts/ directory utilize the subprocess module to chain execution between local Python files.
Evidence: scripts/batch_convert_templates.py uses subprocess.call([sys.executable, str(local_convert), ...]) to run the conversion engine.
Evidence: Wrapper scripts like scripts/convert_to_csv.py and scripts/parse_json.py use subprocess.call to invoke convert_formats.py or parse_formats.py with specific flags.

test-case-reviewer