The Agent Skills Directory

[SAFE]: The skill provides transparent mathematical modeling for noise, dust, and vibration impacts. No obfuscation, hardcoded credentials, or malicious prompt instructions were identified across the 32 analyzed files.
[EXTERNAL_DOWNLOADS]: Documentation in the README indicates no external dependencies; however, the validate_injurious.py utility imports the jsonschema library. This is a standard industry practice for input validation and does not pose a security risk in this context.
[COMMAND_EXECUTION]: The skill contains no calls to subprocesses, shell commands, or dynamic execution sinks (like eval or exec). Operations are limited to arithmetic calculations and JSON file input/output.
[INDIRECT_PROMPT_INJECTION]: The skill presents a data ingestion surface through JSON input files.
Ingestion points: injurious_affection_calculator.py and validate_injurious.py read user-supplied JSON files.
Boundary markers: The injurious_affection_input_schema.json defines a strict schema for all inputs, which serves as a structural boundary.
Capability inventory: No dangerous capabilities (network, shell, or file-write outside of results) were detected in any script.
Sanitization: validate_injurious.py provides robust sanitization through schema validation and explicit type conversion for all input fields.

injurious-affection-assessment