The Agent Skills Directory

[NO_CODE]: The skill is entirely composed of Markdown, JSON, and HTML knowledge files. It does not contain any executable scripts (Python, JavaScript), binaries, or shell commands that could be leveraged for malicious activity.
[SAFE]: The role-playing instructions in SKILL.md are designed to constrain the agent's behavior within a specific medical historical context rather than bypassing global safety filters. They include explicit instructions to avoid medical diagnosis, to never fabricate information not found in the knowledge base, and to remind users to seek professional medical attention for emergencies.
[SAFE]: Data retrieval is restricted to local markdown files provided within the skill package. There are no network operations or external data fetching mechanisms that could lead to exfiltration.
[SAFE]: Installation instructions in the README.md utilize standard development workflows (Git clone, specific skill registries) and do not involve high-risk runtime execution patterns.

like-perspective