resonate-recursive-fan-out-pattern-python
Pass
Audited by Gen Agent Trust Hub on Aug 10, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The
crawlfunction provides a surface for indirect prompt injection by fetching content from external URLs and using extracted data to influence subsequent agent actions. - Ingestion points: The
crawlfunction inSKILL.mdingests data viafetch_page(url), where the URL is derived from previously crawled links. - Boundary markers: Not present. There are no instructions or delimiters provided to prevent the agent from following malicious instructions found within the fetched content.
- Capability inventory: The skill utilizes
ctx.rpcandctx.runto dispatch tasks, which allows external content to influence the execution graph of the agent across worker groups. - Sanitization: No sanitization or URL validation is performed on the links extracted from external pages before they are passed to the next stage of the workflow.
- [SAFE]: The skill utilizes the
resonatePython package, which is a resource provided by the vendor, for managing durable promises and task distribution.
Audit Metadata