content-research-brief
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from the open web, including news articles, blog posts, and social media discussions (Reddit, LinkedIn).
- Ingestion points: The
web_searchandweb_fetchtools are used extensively in Step 1 and Step 2 to retrieve external content. - Boundary markers: The instructions lack explicit delimitation or instructions for the agent to ignore potential instructions embedded within the fetched articles or social media posts.
- Capability inventory: The skill uses the fetched data to synthesize content angles and research briefs. It does not appear to use the data for high-risk actions like executing code, writing to the file system, or modifying environment variables.
- Sanitization: There are no mentioned filters, sanitization protocols, or validation steps for the content fetched from external URLs before it is interpolated into the agent's context for synthesis.
Audit Metadata