frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Show HN: WebMCP Registry of Websites That Agents Can Use

https://www.wmcp.ai/
1•jusquan•29s ago•0 comments

Show HN: Groupicorn directory of IOP group therapy programs

1•cs1996•31s ago•0 comments

Ancient DNA reveals pervasive directional selection across West Eurasia

https://www.nature.com/articles/s41586-026-10358-1
1•gmays•1m ago•0 comments

Code similarity detection using Tree-sitter

https://github.com/dsummersl/treepeat
1•91awebsi•2m ago•1 comments

ES Archive – a new memory system for AI, running natively on Mac

https://github.com/apocryphx/ES-Archive
1•ApocryphX•2m ago•0 comments

Claude Opus 5.5: Intelligence, Performance and Price Analysis

https://artificialanalysis.ai/models/claude-opus-5-5
1•theanonymousone•4m ago•0 comments

Barbara Mazzolai Wants to Build a New Field of Robotics

https://spectrum.ieee.org/sustainability-robotics-barbara-mazzolai
1•SamuraiLion•4m ago•0 comments

Siri AI Settlement Website Now Live

https://www.macrumors.com/2026/09/20/siri-ai-settlement-website-now-live/
1•JumpCrisscross•5m ago•0 comments

Xeno-Interpretability: Investigating the Alien Minds of LLMs

https://arxiv.org/abs/2609.20408
1•potent_latent•5m ago•0 comments

Aging may be a program, not a breakdown

https://www.quantamagazine.org/why-aging-may-be-a-program-not-a-breakdown-20260814/
5•elo2000•7m ago•0 comments

We Rebuilt Jev's API on an Open Model and Used It to Play Doom

https://blocks.ai/blog/jev-open-model-doom
2•stephenblum•8m ago•1 comments

How do you best protect against copycats?

1•thereisnotry•8m ago•0 comments

Tables and Strings in COBOL - Big Data like it's 1985

https://www.datagubbe.se/cobtab/
1•rbanffy•9m ago•0 comments

Show HN: graf (1000x faster graphify in Rust)

https://github.com/ctxrs/graf
1•ripped_britches•9m ago•0 comments

GameStop shares up 30% since earnings; CEO, other board members disclose buys

https://finance.yahoo.com/markets/article/gamestop-stock-jumps-as-ceo-ryan-cohen-buys-26-million-...
1•frmersdog•9m ago•0 comments

Armenia on track to become one of leading hubs for US AI infrastructure

https://armenpress.am/en/article/1261009
1•melicerte•10m ago•1 comments

Apple Developing New Fitness Tracker Aimed at Rivaling Whoop

https://www.bloomberg.com/news/articles/2026-09-22/apple-is-developing-new-fitness-tracker-aimed-...
2•elo2000•11m ago•0 comments

A framework for frontier AI and the dawning of a new age

https://institute.deepmind.com/essays/a-framework-for-frontier-ai-and-the-dawning-of-a-new-age/
2•LyalinDotCom•11m ago•0 comments

Ask HN: Did your side projects ever become something big enough to sustain you?

2•learner_yearner•12m ago•1 comments

18-year-old thought she'd take a gap year. Instead, she joined the S.F. Symphony

https://www.sfchronicle.com/entertainment/classical/article/sf-symphony-starla-breshears-22362042...
1•TMWNN•12m ago•0 comments

OpenRouter Batch API: half-price inference by bundling requests

https://openrouter.ai/blog/announcements/batch-api/
1•porridgeraisin•14m ago•0 comments

'Coding Kids' by Natasha Singer: An Education for the Future That Hasn't Worked

https://www.wsj.com/arts-culture/books/coding-kids-review-an-education-for-the-future-af7fa2a6
1•theanonymousone•14m ago•0 comments

The most difficult daily word game

https://un-scrabbled.com
1•jakespider•14m ago•0 comments

ProudSend – cold outreach drafts that cite the line they came from

https://proudsend.com
1•heypetar•15m ago•0 comments

Show HN: I gave Jev my forum database so he could find the most relevant

https://forumbacklinks.org/demo/setup
1•bigmuzzy•15m ago•0 comments

Go-System-One: Jev-Like Results in Pure Go and SIMD/PTX

https://github.com/rcarmo/go-system-one
1•rcarmo•16m ago•1 comments

Benchmarks are more broken than we could have imagined

https://www.horizonanalyticslabs.com/research/public-benchmark-dataset-audit
3•chillacy•16m ago•1 comments

AI Is Not the End of the World

https://www.everythingisbullshit.blog/p/ai-is-not-the-end-of-the-world
1•swolpers•16m ago•0 comments

Automated optimization of a molecular simulation program

https://shishir-iyer.medium.com/automated-optimization-of-a-molecular-simulation-program-4a28a2bc...
1•shishir03•17m ago•0 comments

RetentionVolt MCP – Giving AI agents video editing references

https://retentionvolt.com
1•AndreaBarretta•18m ago•0 comments
Open in hackernews

Research Harness from First Principles

https://edotenv.com/blog/oh-my-quant-harness
1•RuiWang0811•46m ago

Comments

RuiWang0811•46m ago
TLDR: we built Oh My Quant, a harness designed for research from first principles -> fast & cheap experiment running, monitoring and storing. Our harness with Opus is top 1 on MLE leaderboard.

The point is to let an agent spawn, run, and track a lot of hypotheses and experiments in parallel without confusing itself.

Most harnesses right now are for coding, which is fine for SWE but not ideal for research. Good research means shipping many hypotheses, experiments, and validations, instead of static PRs. We used Codex and Claude Code for a while, but they have poor research 'taste' and are bad at running and comparing multiple experiments.

We ended up designing Oh My Quant from scratch for our own research. The harness is built around hypothesis → experiment → evidence, with a few simple principles: no hardcoded workflows, main agent decides which subagents to spawn; cheap and fast experiment design and tracking; harness handles all tedious things like parsing experiment metadata, run and monitor experiment, store result metrics - and leaves the agent's entire context window free to do the intelligent judgement calls for research design.