auto-research
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and process content from external sources (web search and ChatGPT), which introduces a vulnerability surface where malicious instructions embedded in remote content could attempt to influence the agent's logic.
- Ingestion points: External data enters the agent context via the
playwrightbrowser session andWebSearch/WebFetchtools described in Step 2 of the workflow. - Boundary markers: The skill implements strong manual boundary markers by requiring the agent to distill findings into options (Step 3) and wait for an explicit Pick/Approval step (Step 4) before any code implementation occurs.
- Capability inventory: The skill uses
playwrightfor browser automation and has the capability to modify the local workspace (implementing code) following the research phase. - Sanitization: While the skill mandates redaction of outgoing sensitive information, it relies on the user review of the presented findings as the primary defense against malicious instructions contained within the ingested web data.
Audit Metadata