speculative-decoding
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [SAFE] (SAFE): Comprehensive analysis of the provided documentation and code templates revealed no evidence of malicious behavior. The skill is purely informational, focusing on inference speedup methods.
- [EXTERNAL_DOWNLOADS] (SAFE): The skill documentation references downloading machine learning models from Hugging Face and cloning open-source research repositories from GitHub (hao-ai-lab/LookaheadDecoding and FasterDecoding/Medusa). These references are standard within the technical domain and originate from established research entities.
Audit Metadata