dflash-mlx-speculative-decoding
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs users to install the
dflash-mlxpackage from PyPI and references model weights from Hugging Face (e.g.,Qwen/Qwen3.5-9B,z-lab/Qwen3.5-9B-DFlash). These are standard operations for an AI inference skill. - [COMMAND_EXECUTION]: The skill provides CLI commands for text generation (
dflash), running an inference server (dflash-serve), and benchmarking (dflash-benchmark). These tools are used for their intended purpose within the user's local environment. - [DATA_EXPOSURE]: The
dflash-servecommand starts a local server on port 8000. While it defaults to no authentication (api_key="not-needed"), this is a standard configuration for local development servers intended for use within a private network or on a local machine.
Audit Metadata