Open, reproducible CLI experience evaluation for humans and AI agents.
CLI Rubric is an open-source project for measuring whether a command-line interface is usable, good, or preferred. Human and agent experience are scored separately and never merged.
Important
CLI Rubric is in its founding phase. There is no official rubric, score, badge, certification, or leaderboard yet. Any result produced before a validated rubric release must be labeled experimental.
Read the founding vision.
Review the experimental Rubric v0 design.
- Evidence before scores.
- Human Score and Agent Score stay separate.
- Official results are versioned, inspectable, and reproducible.
- The same public rules apply to Lathe and every other CLI.
- Improving against the public rubric is the point.
- Validation and independent reproduction earn credibility.
The scoring method is not finished. Implementation and packaging decisions remain open until the method is defined.