rk-ai
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted data from research papers and datasets, creating a surface for indirect prompt injection.
- Ingestion points: Reads active papers, scientific notes, and data definitions from the working directory (SKILL.md, references/leakage-and-splits.md).
- Boundary markers: The skill does not define specific delimiters or instructions for the agent to separate user-provided data from its own core instructions.
- Capability inventory: The skill is designed to read local project files and write evaluation records and notes (assets/ml-eval-record.md).
- Sanitization: No explicit sanitization or filtering of external content is described in the procedures.
Audit Metadata