The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were identified. The skill follows best practices for structured document generation in a medical research context.\n- [SAFE]: Strong anti-hallucination controls are implemented, including strict requirements for verified citations (DOIs/PMIDs) and prohibitions against inventing clinical definitions or references from memory.\n- [SAFE]: The skill maintains security boundaries by using explicit TODO markers for institutional information, preventing the agent from generating potentially unauthorized or sensitive administrative content.\n- [SAFE]: Tool usage (Bash, Write, Read) is appropriately scoped to local file management and integration with project artifacts within a research pipeline, with no signs of malicious exfiltration or system modification.

write-protocol