evalite  by mattpocock

TypeScript testing framework for LLM apps

Created 10 months ago
864 stars

Top 41.5% on SourcePulse

GitHubView on GitHub
Project Summary

Evalite is a TypeScript testing framework for LLM-powered applications, designed for developers to efficiently validate their AI-driven features without requiring API keys for core functionality. It offers a streamlined approach to testing, enabling rapid iteration and reliable deployment of LLM integrations.

How It Works

Evalite leverages a local testing paradigm, allowing developers to run tests against their LLM applications directly. The framework focuses on type safety and provides a development server for real-time feedback, facilitating a robust testing workflow.

Quick Start & Requirements

  • Install: pnpm install (in root)
  • Run: pnpm run dev
  • Prerequisites: Node.js, pnpm. An OpenAI API key is required for the example application (packages/example/.env).
  • Setup: Requires running pnpm build in the root and npm link within packages/evalite for global command access.
  • Docs: View the docs
  • Discord: Join the Discord

Highlighted Details

  • TypeScript-first testing framework for LLM applications.
  • Local development server for real-time feedback.
  • Focus on type safety and efficient testing workflows.

Maintenance & Community

  • Community support available via Discord.
  • Contribution guidelines are provided.

Licensing & Compatibility

  • License: Not specified in the README.
  • Compatibility: Designed for TypeScript projects.

Limitations & Caveats

The example application requires an OpenAI API key, and setting up the global evalite command involves an extra npm link step. The licensing is not explicitly stated, which may impact commercial use.

Health Check
Last Commit

3 weeks ago

Responsiveness

1 day

Pull Requests (30d)
4
Issues (30d)
1
Star History
82 stars in the last 30 days

Explore Similar Projects

Starred by Chip Huyen Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems"), Meng Zhang Meng Zhang(Cofounder of TabbyML), and
3 more.

qodo-cover by qodo-ai

0.2%
5k
CLI tool for AI-powered test generation and code coverage enhancement
Created 1 year ago
Updated 2 months ago
Feedback? Help us improve.