confident-ai/deepeval

DeepEval for Typescript

开放

#2,734 创建于 2026年6月9日

 (12 条评论) (2 个反应) (1 位负责人)Python (1,677 个派生)auto 404
documentationenhancementhelp wanted

仓库指标

星标
 (16,939 个星标)
PR 合并指标
 (平均合并 4天 16小时) (30 天内合并 26 个 PR)

描述

Hi folks - we recently released our plans for Typescript support, and we could borrow some help if some is available :)

The current /typescript folder already contains and acts as a client wrapper to Confident AI, and currently already supports tracing.

Our goal is to make local evals, simulation, and synthetic data generation, available in Typescript by July 1st.

We're leaving out benchmarks on purpose because we don't believe that's what Typescript users want.

Here are the initial TODOs:

  • Support local evaluation models
  • Support for complete metrics parity, with shared templates
  • Support for unit test based testing workflows (similar to evals iterator, evaluate function, and deepeval test run)

Post July 1st:

  • Support for complete feature parity with regards to synthetic data generation
  • Support for complete feature parity with regards to simulation
  • Support for prompt optimization

If you're interested, please comment below.

贡献者指南