protocal-agent

This skill is coherent for generating lab protocol plans, but consider the following before installing or running it: - Privacy: the skill will read .md and .docx files in your project root and may incorporate that content into outputs and web queries — avoid placing private or sensitive data (e.g., patient identifiers, proprietary protocols, or credentials) in the project root. - Data exfiltration risk: web searches and included references could cause the agent to send protocol details (including local sample descriptions) to external search engines. If that is a concern, do not allow internet access or sanitize inputs before requesting a plan. - Biological safety: the skill produces wet-lab instructions (e.g., CRISPR, RNA extraction). Treat generated plans as advisory only — do not execute without review by qualified personnel, institutional approvals (IACUC/IRB/IBC if applicable), and compliance with local EHS policies. - Verification: catalog numbers, reagent specifics, and critical parameters should be independently verified against vendor documentation and institutional SOPs before use. If you want greater safety/privacy, ask the skill author to limit local-file access, to avoid including verbatim local data in web queries, or to run in an environment without outbound network access.