discover
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill leverages sub-agents with
WebSearchandWebFetchto gather information from external, untrusted sources such as issue trackers, documentation, and user complaints. - Ingestion points: External data enters the agent context through the use of web search and fetching tools during the research phase described in Step 3 of the Process.
- Boundary markers: The instructions state to "Keep raw research out of the main context," which acts as a partial boundary, but there are no explicit instructions for the agent to disregard malicious directives embedded within the fetched content.
- Capability inventory: The skill has the capability to write the processed information to the local filesystem at the path
docs/<feature>/research.md. - Sanitization: There is no mention of sanitizing, escaping, or validating the content retrieved from the web before it is summarized or written to the documentation file.
Audit Metadata