quaid evals
About

Quaid Evals

Benchmark pages for a local-first memory system.

What This Is

  • Quaid is local-first agent memory with an MCP server and no cloud dependency.
  • Results are published for transparency so anyone can reproduce the numbers.
  • Every release runs through GitHub Actions before scores land on the site.

Methodology

  • Open corpora only, including FiQA and MSMARCO.
  • No synthetic data in the published benchmark runs.
  • Contributions and issue reports live in the public repo.

Links