![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?v=4&s=144)
Commit activity
Last 12 weeks · 0 commits
Community health
3 of 6 standards met
57
✓README✓License✓Contributing○Code of Conduct○Issue Template○PR Template
Recent PRs & issues
Quiet · Last activity 2mo ago
See all on GitHub →
The AI SDK v7 stable release is out, but evalite's range doesn't include it yet
is there a way to provide global statistics at the end? For example let's say I have a total tokens or time spent generating for each task as metrics. I would like to have an avarage time spent generating and an avarage tokens used _Originally posted by @marcoripa96 in https://github.com/mattpocock/evalite/discussions/341_
Summary: Add MIT license metadata to the published evalite package manifest. Validation: node package metadata assertion npm pack --dry-run --json --ignore-scripts in packages/evalite git diff --check
Recent fixes
With Vitest 3.1+'s plugin hook, it's possible to integrate Evalite directly as a Vite plugin rather than requiring a separate and custom CLI (, ). A could register the reporter, configure test patterns for , and spin up the UI server—all through standard Vite/Vitest primitives. Users would just add the plugin to their existing and run as normal. This would reduce surface area, eliminate the duplicated CLI, and let Evalite compose naturally with existing Vite/Vitest setups.
Structured data for AI agents
Repository: mattpocock/evalite. Description: Evaluate your LLM-powered apps with TypeScript Stars: 1610, Forks: 94. Primary language: TypeScript. Languages: TypeScript (89.8%), MDX (7.9%), CSS (1.1%), JavaScript (0.8%), HTML (0.2%). License: MIT. Homepage: https://www.evalite.dev/ Topics: ai, evals, typescript. Latest release: evalite@0.19.0 (7mo ago). Open PRs: 14, open issues: 45. Last activity: 2mo ago. Community health: 57%. Top contributors: mattpocock, github-actions[bot], christianklotz, DogPawHat, jacobparis, tyom, dschlabach, vojtaholik, cdavis1324, iamladi and others.