The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts (export_openapi_from_java.py and generate_openapi.py) to process Java files. While this involves command execution, it is restricted to internal scripts within the skill's directory and follows predefined templates.
[PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection. It parses Java source files, specifically extracting content from Javadoc comments and annotations (e.g., @BusinessDescriptor, @Schema). Malicious instructions embedded in the target Java code's documentation could theoretically influence the agent's behavior during the extraction phase.
Ingestion points: Java source files provided via the --source argument.
Boundary markers: None explicitly implemented for parsing comments.
Capability inventory: Subprocess execution (Python scripts) and file system writes (OpenAPI JSON output).
Sanitization: None detected for the extracted string literals from Java code before display or processing.
[SAFE]: Static analysis identified a subprocess.run call in tests/test_export_openapi_from_java.py. This is a benign usage for regression testing, as it explicitly calls the Python executable with a list of arguments and does not use a shell environment.

ac-api