interpreting-microbiology-cultures
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is composed entirely of instructional text and markdown tables designed to guide the interpretation of medical pathology data. It does not contain any functional code or scripts.
- [DATA_EXPOSURE_AND_EXFILTRATION]: No sensitive file paths, hardcoded credentials, or network-enabled commands (such as curl or wget) were detected. The skill does not interact with the host system's data.
- [REMOTE_CODE_EXECUTION]: There are no patterns involving the download, installation, or execution of external code or software packages.
- [PROMPT_INJECTION]: The content strictly adheres to the stated purpose of microbiology analysis. No attempts to bypass safety guardrails, override system prompts, or extract internal instructions were found.
- [INDIRECT_PROMPT_INJECTION]: While the skill is designed to process external data (lab reports and clinical notes), it does not possess any capabilities (like file writing or network access) that could be exploited by malicious content within that data. The instructions serve as a structural guide for the agent's reasoning process.
Audit Metadata