research-synthesis
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Processing of Untrusted Data: This skill is designed to ingest and synthesize external data such as interview transcripts, survey results, and support tickets. These sources may contain embedded instructions intended to influence the agent's behavior (indirect prompt injection).
- Ingestion points: Data enters the agent's context through user-provided arguments and connected feedback/analytics tools defined in
SKILL.md. - Boundary markers: The current instructions do not include specific delimiters or directives to the agent to ignore potentially malicious instructions embedded within the research data.
- Capability inventory: The skill primarily generates markdown output, but if used in conjunction with tools like a knowledge base or product analytics, the synthesized output could influence downstream operations.
- Sanitization: There are no explicit instructions for the agent to sanitize or validate the content of the research data before processing.
Audit Metadata