The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were detected in this skill. The skill provides legitimate documentation and automation scripts for healthcare application testing.
[COMMAND_EXECUTION]: The skill provides examples of shell commands using npx jest, jq, and bc. These are standard tools used for automated testing and JSON processing in development environments and do not represent a security risk in this context.
[REMOTE_CODE_EXECUTION]: The instructions include npx jest, which executes a package from the registry. This is standard behavior for a JavaScript testing harness and is used appropriately here.
[INDIRECT_PROMPT_INJECTION]: The skill describes processes for ingesting and reporting on test data.
Ingestion points: Test output files processed via jq (SKILL.md).
Boundary markers: Standard command-line redirection and piping.
Capability inventory: Subprocess execution restricted to local testing tools.
Sanitization: Not applicable as it primarily handles numeric pass/fail counts.
[METADATA_POISONING]: The metadata includes attribution to a healthcare provider and a specific doctor. While the author 'eugene-ee' differs from the 'origin' metadata field, this appears to be a standard attribution for contributed content and does not show signs of deceptive intent to bypass security controls.

healthcare-eval-harness