The Agent Skills Directory

[SAFE]: The skill provides educational content and local utility scripts for assessing telehealth compliance. No network operations, sensitive file access, or command execution patterns were identified in the provided files.
[METADATA_POISONING]: The metadata fields (name, description, author) accurately reflect the content and purpose of the skill. No deceptive or hidden instructions were found in the YAML frontmatter.
[REMOTE_CODE_EXECUTION]: The Python script scripts/process.py only imports standard libraries (json, os, datetime, typing) and does not perform any remote package installation or code downloads.
[DATA_EXFILTRATION]: There are no network-capable functions (e.g., requests, curl, urllib) or hardcoded secrets/credentials in any of the skill's components.
[DYNAMIC_EXECUTION]: The code uses standard dictionary lookups and conditional logic. No usage of eval(), exec(), or dynamic path loading was observed.
[INDIRECT_PROMPT_INJECTION]: While the skill processes data through its scripts, it lacks dangerous capabilities (file writes, network calls, command execution) that could be exploited via malicious input. The script behavior is deterministic and safe.

telehealth-privacy