Feed aggregator

Article URL: https://aembit.io/blog/mcp-oauth-2-1-pkce-and-the-future-of-ai-authorization/

Comments URL: https://news.ycombinator.com/item?id=44043629

Points: 1

# Comments: 0

I've built an open-source implementation of Google DeepMind's AlphaEvolve system called OpenEvolve. It's an evolutionary coding agent that uses LLMs to discover and optimize algorithms through iterative evolution.

Try it out: https://github.com/codelion/openevolve

What is this?

OpenEvolve evolves entire codebases (not just single functions) by leveraging an ensemble of LLMs combined with automated evaluation. It follows the evolutionary approach described in the AlphaEvolve paper but is fully open source and configurable.

I built this because I wanted to experiment with evolutionary code generation and see if I could replicate DeepMind's results. The original system successfully improved Google's data centers and found new mathematical algorithms, but no implementation was released.

How it works

The system has four main components that work together in an evolutionary loop:

1. Program Database: Stores programs and their metrics in a MAP-Elites inspired structure 2. Prompt Sampler: Creates context-rich prompts with past solutions 3. LLM Ensemble: Generates code modifications using multiple models 4. Evaluator Pool: Tests programs and provides feedback metrics

What you can do with it

- Run existing examples to see evolution in action - Define your own problems with custom evaluation functions - Configure LLM backends (works with any OpenAI-compatible API) - Use multiple LLMs in ensemble for better results - Optimize algorithms with multiple objectives

Two examples I've replicated from the AlphaEvolve paper:

- Circle Packing: Evolved from simple geometric patterns to sophisticated mathematical optimization, reaching 99.97% of DeepMind's reported results (2.634 vs 2.635 sum of radii for n=26). - Function Minimization: Transformed a random search into a complete simulated annealing algorithm with cooling schedules and adaptive step sizes.

Technical insights

- Low latency LLMs are critical for rapid generation cycles - Best results using Gemini-Flash-2.0-lite + Gemini-Flash-2.0 as the ensemble - For the circle packing problem, Gemini-Flash-2.0 + Claude-Sonnet-3.7 performed best - Cerebras AI's API provided the fastest inference speeds - Two-phase approach (exploration then exploitation) worked best for complex problems

Getting started (takes < 2 minutes)

# Clone and install git clone https://github.com/codelion/openevolve.git cd openevolve pip install -e .

# Run the function minimization example python openevolve-run.py examples/function_minimization/initial_program.py \ examples/function_minimization/evaluator.py \ --config examples/function_minimization/config.yaml \ --iterations 50

All you need is Python 3.9+ and an API key for an LLM service. Configuration is done through simple YAML files.

I'll be around to answer questions and discuss!

Comments URL: https://news.ycombinator.com/item?id=44043625

Points: 1

# Comments: 0

Article URL: https://www.youtube.com/watch?v=o8NiE3XMPrM

Comments URL: https://news.ycombinator.com/item?id=44043618

Points: 3

# Comments: 0

Article URL: https://www.bloomberg.com/news/articles/2025-05-20/doordash-ends-ai-voice-ordering-product-for-restaurants

Comments URL: https://news.ycombinator.com/item?id=44043609

Points: 1

# Comments: 1

Article URL: https://blog.runreveal.com/introducing-pipelines-in-runreveal/

Comments URL: https://news.ycombinator.com/item?id=44043607

Points: 1

# Comments: 0

The true mistake is not learning from your own mistakes.

My platform helps you avoid that with Mistakes Tracking, Prevention Center, Advanced Analytics, and more ... to learn from your own mistakes. Designed with first principles thinking in mind.

Comments URL: https://news.ycombinator.com/item?id=44043604

Points: 1

# Comments: 0

Article URL: https://anephenix.com/blog/post/introducing-event-emitter

Comments URL: https://news.ycombinator.com/item?id=44043603

Points: 1

# Comments: 0

Article URL: https://www.youtube.com/watch?v=HsAatv08e9U

Comments URL: https://news.ycombinator.com/item?id=44043601

Points: 2

# Comments: 0

Article URL: https://github.com/microsoft/BitNet/blob/main/gpu/README.md

Comments URL: https://news.ycombinator.com/item?id=44043600

Points: 1

# Comments: 0

Article URL: https://www.paolomainardi.com/posts/docker-performance-macos-2025/

Comments URL: https://news.ycombinator.com/item?id=44043596

Points: 1

# Comments: 0

Marketing is one of the hardest parts to figure out for small businesses and solopreneurs.

So we built Markebility, a tool designed to make marketing easier by using AI to walk users through the full marketing journey. The tool helps generate a simple strategy, weekly content plan (focused on social media for now), and generates posts for those channels.

We’re finalizing integrations with the social platforms now. Once that’s done, users will be able to schedule and publish directly from the platform. The goal is to also pull in analytics from those posts so we can improve the strategy over time and recommend better-performing content and channels.

Right now, the strategy, content planner and content generator are live and free to try. We still have a ways to go, but this is the most complete version we've shipped since early MVPs so we're looking for feedback. Particularly on the onboarding flow, clarity of value, or anything that feels off or missing. If you have any ideas on what would make this tool even better, we'd love to hear it.

Thanks for checking it out!

Comments URL: https://news.ycombinator.com/item?id=44043593

Points: 2

# Comments: 0

Article URL: https://countercraft.substack.com/p/newspapers-are-recommending-ai-hallucinated

Comments URL: https://news.ycombinator.com/item?id=44043582

Points: 2

# Comments: 0

Article URL: https://usgraphics.com/public-affairs/bulletins/BT-002

Comments URL: https://news.ycombinator.com/item?id=44043571

Points: 1

# Comments: 0

Article URL: https://deepmind.google/technologies/lyria/

Comments URL: https://news.ycombinator.com/item?id=44043564

Points: 3

# Comments: 0

Article URL: https://www.theregister.com/2025/05/20/adobe_price_hikes/

Comments URL: https://news.ycombinator.com/item?id=44043559

Points: 3

# Comments: 0

Hi HN,

I built Streamix as a fresh take on reactive streams for JavaScript and TypeScript.

It’s inspired by RxJS, but designed to be lighter, simpler, and more ergonomic — especially for developers who just want clean async data pipelines without having to learn a full reactive framework.

Some features:

Fully pull-based (async iterable in, async iterable out)

Composable operators like map, filter, mergeMap, audit, etc.

A coroutine operator that offloads heavy processing to Web Workers

Zero dependencies, tree-shakable, TypeScript-first

Small and fast — ideal for frontends, workers, and reactive logic

I created Streamix because I often wanted RxJS-like power without RxJS-level complexity. Most stream libraries are either too limited or too heavy. Streamix aims for a sweet spot: expressive, composable, and easy to debug.

You can see code examples and usage in the GitHub repo: https://github.com/actioncrew/streamix

I’d love feedback from the HN community — whether it’s feature ideas, API critiques, or just thoughts on the reactive model.

Thanks!

Comments URL: https://news.ycombinator.com/item?id=44043552

Points: 2

# Comments: 0

Article URL: https://www.thetimes.com/us/news-today/article/air-traffic-controller-newark-airport-nj-9zpp3z9ww

Comments URL: https://news.ycombinator.com/item?id=44043549

Points: 1

# Comments: 0

Jobamate helps you auto-apply to jobs with a few clicks. Upload your resume, and it finds matches, tailors cover letters using GPT-4, and even fills out job apps on LinkedIn and Indeed using browser automation

I built this after watching too many friends burn out from applying manually. It uses resume parsing, skill-to-role matching, and dynamic prompt engineering to target real jobs—not just generic spam blasts.

There's a Chrome extension that works in the background as you browse job boards, and a dashboard to track progress and generate custom interview prep. I'm also working on a mobile version and deeper integrations with niche boards.

Would love feedback from anyone job hunting, hiring, or thinking about resume AI.

Comments URL: https://news.ycombinator.com/item?id=44043542

Points: 2

# Comments: 0

Article URL: https://www.macrumors.com/2025/05/20/google-releases-notebooklm-app-for-ios-and-android/

Comments URL: https://news.ycombinator.com/item?id=44043540

Points: 1

# Comments: 0

While 79% of security leaders believe that AI agents will introduce new security and compliance challenges, 80% say AI agents will introduce new security opportunities.