tw-research-lit-review
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources which could contain malicious instructions designed to influence the agent's behavior.
- Ingestion points: The skill uses
WebSearchin Step 3 and Step 4 to fetch academic literature and search results from various databases (Google Scholar, Web of Science, etc.). - Boundary markers: The prompt lacks explicit instruction delimiters or 'ignore embedded instructions' warnings when processing search results, though it does include a generic rigorousness statement.
- Capability inventory: The skill has access to
Bash,Write, andNotiontools, allowing it to modify local files or external database entries based on processed content. - Sanitization: No explicit sanitization or filtering of external content is described before interpolation into the literature review draft or Notion cards.
Audit Metadata