deep-research
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from the web via sub-agents, creating a vulnerability surface where malicious instructions in research sources could influence agent behavior. \n
- Ingestion points: Research findings extracted from external websites via browser tools and stored in workspace markdown files. \n
- Boundary markers: The skill relies on natural language instructions to guide sub-agents but lacks robust technical delimiters or explicit 'ignore instructions' warnings for external content. \n
- Capability inventory: The skill can execute local Python scripts for saturation checks and manage a local HTTP server. \n
- Sanitization: There is no evidence of automated filtering or sanitization of the scraped content before it is processed by the synthesizing agent. \n- [SAFE]: The visualizer dashboard utilizes standard, well-known libraries (ECharts and Marked) sourced from the JSDelivr CDN, which is a trusted service for front-end assets. The visualization server itself is a transparent implementation designed for local reporting and project management, requiring the user to manually initiate the process in a separate terminal.
Audit Metadata