Benchmark unlimited Claude.md files against eachother
Clawmark is a tool that benchmarks unlimited Claude.md files against each other, allowing users to compare performance across different configurations or versions.
Background
- **Clawmark** is an open-source tool for benchmarking Claude.ai project knowledge (`.md`) files — letting users test which version of a prompt, instruction set, or context document performs better at a given task.
- It works by running multiple `.md` files against the same test queries and scoring the results, so users can systematically compare different versions of their Claude project instructions.
- This matters because Claude's "Projects" feature lets you upload custom knowledge files, but there's no built-in way to A/B test or measure which version of those files yields better outputs — Clawmark fills that gap for power users and prompt engineers.
- The repo is by Emiliano Lugo and is aimed at developers and advanced Claude users who want to iterate on project instructions in a data-driven way.