The Agent Skills Directory

[SAFE]: No security issues detected. The skill consists of informational content regarding Human-Computer Interaction principles and Python code examples for usability testing, accessibility auditing, and eye-tracking analysis. These examples use standard libraries and do not perform any sensitive file access, network operations, or privileged command execution.
[INDIRECT_PROMPT_INJECTION]: The skill includes code structures for processing external data, such as usability participant comments and web page data. This represents a potential surface for indirect prompt injection if the skill were to process adversarial input in an active agent environment. However, the provided scripts are educational examples and do not implement any automated or dangerous actions based on that data.
Ingestion points: record_task_completion in SKILL.md (comments field), audit_page in SKILL.md (page_data dictionary).
Boundary markers: None provided in the boilerplate examples.
Capability inventory: Data analysis, report generation, and basic arithmetic. No file system writes or network capabilities are present.
Sanitization: Not applicable to these educational code snippets.

human-computer-interaction