medical-paper-humanizer
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or dangerous command executions were detected. The skill is entirely descriptive and instruction-based, functioning as a prompt template for natural language processing.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data (medical papers).
- Ingestion points: The skill processes arbitrary text provided by the user in the
Example Usageand intended clinical workflows. - Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the provided documentation.
- Capability inventory: The skill has zero access to subprocess calls, file system operations, or network tools. It lacks the
allowed-toolsfield in its frontmatter, meaning it cannot execute actions even if prompted by injected text. - Sanitization: No sanitization of input text is described. However, given the zero-capability environment, the risk of a payload in a medical paper influencing the system beyond the immediate text output is negligible.
Audit Metadata