skills/modular/skills/serve-model/Gen Agent Trust Hub

serve-model

Pass

Audited by Gen Agent Trust Hub on Aug 5, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides instructions to install the pixi and uv package managers via shell-piped installation scripts (e.g., curl -fsSL https://pixi.sh/install.sh | sh). These are official installation methods for well-known developer tools.\n- [EXTERNAL_DOWNLOADS]: The skill facilitates downloading of environment components and model configuration from trusted sources, including Modular's official package indices (conda.modular.com, whl.modular.com) and Hugging Face (huggingface.co).\n- [COMMAND_EXECUTION]: A bundled Python script scripts/suggest_serve_command.py is used to statically analyze model architecture and configuration files to recommend optimal server settings. This is a local execution of a skill-provided utility.\n- [REMOTE_CODE_EXECUTION]: The documentation explains the use of the --trust-remote-code flag for models with custom logic, including explicit warnings about the risks of running code from untrusted repositories.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 5, 2026, 12:30 PM
Security Audit — agent-trust-hub — serve-model