EvalView - AI Agent Testing

Version updated for https://github.com/hidai25/eval-view to version v0.1.5.

  • This action is used across all versions by ? repositories.

Go to the GitHub Marketplace to find the latest changes.

Release notes

What’s New

Statistical Pass/Fail System

  • Variance-aware testing - Run tests multiple times to get statistically significant results
  • Confidence levels - Configure how confident you want to be in pass/fail decisions
  • CLI integration - New --runs flag to run tests multiple times
# Run each test 5 times for statistical analysis
evalview run --runs 5

LangGraph Adapter Fix

  • Fixed adapter compatibility issues for better LangGraph integration

Config-Free Runs

  • Run evalview run without requiring a config file
  • Automatically discovers test cases in the current directory

Templates

  • Added test case templates for common evaluation patterns
  • Quick-start templates for tool calling, RAG, and multi-turn scenarios

Node SDK License Fix

  • Fixed license mismatch - now correctly uses Apache 2.0

Documentation Improvements

  • Added FAQ section and comparison table to README
  • Added “Run examples directly” section
  • Added design partners section
  • Improved README structure for better clarity

Full Changelog

https://github.com/hidai25/eval-view/compare/v0.1.4...v0.1.5