The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of various CLI commands for model benchmarking, including accuracy tests and performance monitoring using the evalscope tool.
[COMMAND_EXECUTION]: Includes instructions to host a local dashboard using evalscope service --host 0.0.0.0, which binds the service to all available network interfaces on the host.
[EXTERNAL_DOWNLOADS]: Recommends the installation of the evalscope package and its optional specialized dependencies (e.g., evalscope[perf], evalscope[rag]) from the Python Package Index (PyPI).
[SAFE]: Proactively recommends the use of a Docker-based sandbox when executing benchmarks that require running model-generated code (such as HumanEval or MBPP) to ensure environment isolation and protect the host system.

evalscope-cli