TERMS.md

# Terms

CompText Benchmark 0.1.x is an experimental developer/research plugin for reproducible benchmarking. Benchmark outputs are evidence for the configured fixtures and execution conditions, not universal performance claims.

Users are responsible for authorization to execute workloads and for complying with the terms of any model provider, repository host, or runner they choose.

The plugin is provided without warranty for experimental use. Failed runs, regressions, and missing evidence must remain visible and must not be rewritten into successful claims.