rwkv-architecture
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill's installation guide fetches dependencies from official and well-known sources in the machine learning ecosystem.
- [DYNAMIC_EXECUTION]: Technical documentation demonstrates the use of standard model state serialization patterns.
- [INDIRECT_PROMPT_INJECTION]: The skill contains workflows for processing large documents and user-provided prompts.
- [INGESTION_POINTS]: SKILL.md describes methods for processing user prompts and external documents via load_document().
- [CAPABILITY_INVENTORY]: The model capabilities are limited to text inference (model.forward) and do not involve high-risk operations such as file system modifications or network exfiltration based on the processed content.
- [SANITIZATION]: While standard for model guides, there is no specific data sanitization mentioned; however, the lack of dangerous capabilities mitigates the risk.
Audit Metadata