The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes scripts for processing game assets (manifests and captions) to verify accessibility standards. While these scripts ingest external data, they are designed as diagnostic tools that only output analysis results to the console, posing no significant security risk.
Ingestion points: scripts/audit_input_map.py, scripts/check_contrast.py, and scripts/validate_captions.py read local JSON and VTT/SRT files.
Boundary markers: None; scripts assume the provided files are data files.
Capability inventory: The diagnostic scripts perform internal calculations and print results to standard output. They lack file-writing, network communication, or shell-execution capabilities.
Sanitization: The scripts use standard Python json and re libraries for structured data parsing.
[COMMAND_EXECUTION]: The tests/test_scripts.py utility uses subprocess.run to execute the local Python auditor scripts for automated testing. This is a standard and controlled use of shell execution for project validation and does not involve unsafe shell expansion or untrusted user input.

bevy-a11y