The Agent Skills Directory

[SAFE]: The skill defines a specialized persona for healthcare guidance, specifically medication therapy management and pharmacokinetic dosing. The analysis confirms that the skill does not request sensitive permissions, perform network exfiltration, or include obfuscated malicious payloads. External links target reputable medical organizations and official documentation, posing no risk to the execution environment.- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted medical data (medication lists, histories) from users, creating a surface for indirect prompt injection. However, the skill lacks active capabilities (tool use, network access, or file modification), limiting the impact of such an attack to the generation of incorrect responses. \n\n * Ingestion points: Patient medication lists and clinical history provided in user prompts (SKILL.md, references/workflow.md). \n * Boundary markers: Not present in the current instructions. \n * Capability inventory: No tools, network operations, or file-system modifications are defined within the skill. \n * Sanitization: No specific input validation or sanitization logic is implemented for external data.

clinical-pharmacist