testing

Introduces the Pro Agent Builders series, focusing on recursive auto-improvement (RAI) where a coding agent runs hundreds of probes overnight to automatically improve another agent's performance.

SQLite developers share their reliability experience, derived from the DO-178B avionics standard, ensuring code quality through 100% MCDC test coverage, emphasizing the philosophy: 'if it hasn't been tested, it doesn't work.'

LangWatch released 'scenario', a tool for testing AI agents using conversation scenarios, available on GitHub.

A Google Testing blog post offering guidance on how to respond to code review comments, emphasizing the importance of adding context when it helps clarify decisions and rationale.

pytest-leak-finder is a pytest plugin that uses binary search to identify which earlier test causes a later test to fail when run together.

pgtestdb is a Go/Postgres testing package that uses Postgres template databases to quickly clone test databases, achieving setup times comparable to schema-based methods while enabling full end-to-end testing.

A blog post presenting a universal approach to mocking in testing by using continuations to capture effects, allowing tests to verify correct arguments and order of calls without performing the effects. The technique is demonstrated in the Gleam programming language.

Addy Osmani discusses that with AI agents generating more code than humans can review, software quality must be enforced through constraints like tests and deterministic checks rather than code review itself.

Supapool is a CLI tool that provides isolated, ephemeral Supabase instances for each coding agent or CI run, spinning up in ~400 ms to enable parallel, production-like testing without interference.

Explores why AI agents that perform perfectly on test cases often fail in real-world conversations, highlighting issues like distribution shift and overfitting.

Kent C. Dodds is seeking help for final testing of his new video sharing product before launch.

Xiaomi highlights the safety and durability of the SkyNomad N90 Max, featuring ultra-long sliding rails made from 7-series high-strength aluminum alloy with high locking strength and reliability after 15,000 test cycles.

Balder is a Python test framework that allows writing reusable scenario-based tests that can be run on different devices or technologies without rewriting code.

CreditGenie uses LangSmith to debug thousands of agent traces and generate targeted test questions from production data.

DeepEval is an open-source Python framework for evaluating LLM applications, offering ready-made metrics, Pytest-style workflows, and CI/CD support.

This article discusses how human testers who evaluate AI systems for dangers are struggling to keep up with the rapid pace of AI development, highlighting growing concerns about safety oversight.

This guide explains how to evaluate the quality of AI-generated code using tests, golden datasets, reliability checks, and human review. It provides a practical workflow for catching regressions and shipping AI-assisted code with more confidence.

An analysis investigates whether AI labs are optimizing their models for the popular 'pelican riding a bicycle' SVG benchmark, testing seven frontier models across 48 prompts with varied animals and vehicles, finding no strong evidence of overfitting.

This article introduces the concept of prefactoring - cleaning up code before adding a new feature to make integration cleaner and reduce technical debt.

Claude Code Desktop now includes an iOS Simulator pane for testing and interacting with iOS apps directly within the conversation. The pane streams the device screen live and allows both Claude and the user to control the simulator.