research
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is a set of guidelines for technical research and contains no scripts, binaries, or commands. It does not perform unauthorized network operations or access sensitive system credentials.
- [PROMPT_INJECTION]: The workflow involves processing results from the live web, creating a surface for indirect prompt injection. 1. Ingestion points: Data retrieved from web searches and live source URLs (SKILL.md). 2. Boundary markers: No delimiters or instructions to ignore embedded commands are specified for ingested content. 3. Capability inventory: The skill provides recommendations and does not invoke tools for file modification, system configuration, or remote execution. 4. Sanitization: Relies on multi-source agreement and human-like logical review rather than automated technical sanitization.
Audit Metadata