We submitted Needle 2 here a few weeks ago, and the feedback in the discussion thread was incredibly valuable, thanks! Thanks to all that feedback, we’ve been able to move quickly to release Needle 3 and I'd love to hear what you think again.
The key features:
1) Automation (tool calls & structured JSON output): Needle still doesn't chat by design, its quite challenging to pack general capacity into such small models, so we focus on tool calls and structured JSON. If no tool you declared fits the request, you get an empty list back (note for when playing with the demo).
2) Intelligence Laddering: Every layer (2 to 20) is a deployable subnetwork, so one set of weights, 25 to 121 million parameters at 2-bit, shipping as 8-29MB binaries. On a Raspberry Pi 5 it decodes at up to 4k tokens/sec and prefills at up to 10k.
3) Monarch Hadamard MLP: replaces the dense FFN with three learnable Walsh-Hadamard-initialized Kronecker (Monarch) factor pairs interleaved with per-channel diagonal scales, fixed permutations, a SiLU nonlinearity, and a rank-8 input-conditioned gate, so each token gets a fully mixed nonlinear transform of its d_model channels at O(d√d) parameters and compute instead of the O(d²) a dense 4x-expansion MLP would cost.
4) Performance: On Mobile Actions (phone commands, scored on the exact call) the 20-layer model gets 86.0 through the shipped 2-bit binary; LFM2.5 1.2B is at 82.4, Qwen3.5 0.8B at 76.0, Apple's on-device model at 57.6, all at f16. More results on the link, we do not win everywhere ofc.
5) Multilingual: Needle 3 now supports English, French, Spanish, German, Dutch, Italian, Polish, with more languages coming.
6) Finetuning: You can achieve DeepSeek v4 Flash grade performance on a narrow task with just 4L, stress on "narrow task", we found that production users often prefer tuning before production.
7) Triggers: Grounding is a common challenge for tool call, at least for Needle 2, so we added support case-insensitive regular expressions matched against each request to gate false negatives.
8) Confidence: Every response also carries a calibrated confidence score, the minimum of a judgement on the finished call and its decode probability. Act above your threshold, show the call and ask below it, or escalate to a bigger model.
9) Supported Platforms: macOS, Linux on x86-64, ARM64, ARMv7, RISC-V and MIPS32, Windows x64 and ARM, Android, iOS, watchOS, tvOS, the browser as WebAssembly, and a WASI component.
Thanks for reading and as always, thoughts appreciated!","time":1789690304,"title":"Show HN: Cactus Needle 3: 8-29MB automation models can match DeepSeek V4 Flash","type":"story","url":"https://cactuscompute.com/needle"},{"by":"synack","descendants":61,"id":49757050,"kids":[49758719,49759027,49757776,49762386,49757446,49757457,49757565,49758231],"score":167,"time":1789750458,"title":"Photon-Emission-Guided Laser Fault Injection Enables RP2350 Secure Debug","type":"story","url":"https://donjon.ledger.com/blog/rp2350-secure-debug-laser-fault-injection/"},{"by":"ilreb","descendants":250,"id":49752041,"kids":[49752839,49755873,49756727,49752470,49752419,49752168,49753778,49752791,49757653,49754851,49755172,49752729,49752664,49755532,49753794,49752509,49754286,49754154,49758874,49759638,49752377,49752582,49752754,49758030,49752288,49754012,49753181,49757064,49755314,49754325,49752250,49753358,49753029,49753575,49752575,49762288,49754730,49753886,49752261,49752474,49752565,49759548,49753268,49756689,49752527,49752414,49752346,49752227,49753806,49752349,49752501,49761282,49754736,49760076],"score":583,"time":1789724542,"title":"OpenJev","type":"story","url":"https://openjev.com/"},{"by":"rochansinha","descendants":12,"id":49758615,"kids":[49760096,49759646,49759691,49759609,49762336,49760777,49762520],"score":76,"time":1789757735,"title":"Cache-to-Cache: Direct Semantic Communication Between LLMs (2025)","type":"story","url":"https://arxiv.org/abs/2510.03215"},{"by":"grigolin","descendants":5,"id":49744036,"kids":[49762137,49762278,49762457,49761918],"score":33,"time":1789666741,"title":"The Farnese letter","type":"story","url":"https://simonklee.dk/farnese-letter"},{"by":"torutofu","descendants":1,"id":49726773,"kids":[49763387],"score":10,"time":1789565430,"title":"Goroutine Leak Profiles","type":"story","url":"https://go.dev/blog/goroutine-leak-profiles"},{"by":"so-cal-schemer","descendants":2,"id":49733523,"kids":[49733524],"score":19,"time":1789595384,"title":"LispBM is a concurrent Lisp for microcontrollers with message passing","type":"story","url":"https://www.lispbm.com/"},{"by":"gone35","descendants":15,"id":49731413,"kids":[49761094,49761651,49760626,49760575,49762425],"score":44,"time":1789585269,"title":"Cyclomatic Complexity in C#","type":"story","url":"https://blog.ndepend.com/understanding-cyclomatic-complexity/"},{"by":"datadrivenangel","descendants":206,"id":49760187,"kids":[49762619,49763340,49760499,49760389,49760745,49760447,49762820,49761888,49760738,49760774,49760414,49760765,49760444,49760606,49761305,49761933,49760323,49760494,49761867,49762073,49761158,49762018,49760485,49763228,49761842,49760367,49761610,49763111,49762931,49761810,49760706,49760700,49760405,49761625,49761059,49760391,49760560,49761383,49760319,49760885,49761090,49761468,49761166,49760890,49760785,49761251,49760399,49760733,49762087,49761186,49761906,49761865,49761681,49760255,49760922,49760764,49760652,49760781,49760637,49760418,49760417,49760717,49762822,49761050,49761741,49760449,49762627,49761205,49760533,49760380,49761141,49761629,49760587,49760510],"score":576,"time":1789765232,"title":"Claude Code now reads AGENTS.md if there is no Claude.md","type":"story","url":"https://code.claude.com/docs/en/changelog"},{"by":"nashashmi","descendants":199,"id":49749369,"kids":[49761996,49759018,49760567,49760438,49759026,49753440,49759295,49762625,49759674,49751320,49760845,49760670,49760531,49761666,49763316,49750332,49761813,49763200,49763024,49760033,49763019,49759451,49754605,49762780,49758957,49762656,49761437,49750556,49760836,49759545,49760561,49760107,49762654,49761596,49751005,49751031,49750171,49759875,49760826,49754432,49760099,49751344,49760122,49750023,49761026,49755560,49750013,49762297,49759135,49749999,49762505,49750949,49751114,49751139,49760039,49750080,49759607,49751263,49759517,49750085,49759942,49760326,49760075,49762908,49750718,49751026,49752468,49763186,49761541,49750967,49759001],"score":218,"time":1789696825,"title":"Minimal Phone 2","type":"story","url":"https://minimalcompany.com/"},{"by":"harisingh1612","descendants":32,"id":49762099,"kids":[49763314,49763292,49763074,49762758,49762946,49762533,49762979,49762869,49762800,49762582,49762551,49762267],"score":26,"time":1789777834,"title":"Show HN: LiveWorld – Every 24/7 YouTube live camera on one globe","type":"story","url":"https://liveworld.info/"},{"by":"yogthos","descendants":9,"id":49761840,"kids":[49763232,49762829,49762968,49762452],"score":74,"time":1789775682,"title":"Alibaba open-sources AI model that can detect cancer and nearly 150 conditions","type":"story","url":"https://www.scmp.com/tech/big-tech/article/3368055/alibaba-open-sources-medical-ai-model-can-detect-cancer-and-nearly-150-conditions"},{"by":"ent101","descendants":3,"id":49762835,"kids":[49763395,49763364,49763350],"score":27,"time":1789786224,"title":"Flock Offers Employees Buyouts as Customers Flee","type":"story","url":"https://www.wired.com/story/flock-is-offering-voluntary-buyouts-to-employees/"},{"by":"succinct_ideas","descendants":39,"id":49749724,"kids":[49762497,49761492,49759728,49762181,49758930,49751159,49763062,49762095,49760868,49762873,49760059,49750672,49763252,49759343,49756576,49762638,49762575,49749725,49756582],"score":122,"time":1789700208,"title":"Warez: The Infrastructure and Aesthetics of Piracy (2021)","type":"story","url":"https://archive.org/details/b904a8eb-9c98-4bb1-bf25-3cb9d075b157"},{"by":"csmantle","descendants":94,"id":49750694,"kids":[49759039,49754215,49754510,49751266,49753547,49751804,49755357,49753589,49756783,49752889,49752964,49757527,49762488,49754017,49753494,49755905,49753009,49754746,49754345,49753085,49753030,49753023,49753600,49753643,49753305,49754148,49758965,49753892,49751438,49752903,49755041,49754987,49754360,49752901,49757393,49754912,49753523,49753657,49758290,49754202,49753049,49752672,49753161,49750738,49753889,49753729],"score":273,"time":1789711877,"title":"Inside ZCode: Silently uploading your Git history to the cloud","type":"story","url":"https://blog.ferstar.org/en/posts/zcode-silent-workspace-snapshot-upload/"},{"by":"JumpCrisscross","descendants":52,"id":49746626,"kids":[49761569,49759243,49760887,49749198,49761222,49760024,49750200,49750193,49763034],"score":184,"time":1789679659,"title":"How SpaceX streamlined the Raptor engine","type":"story","url":"https://www.construction-physics.com/p/how-spacex-streamlined-the-raptor"},{"by":"tomjakubowski","descendants":22,"id":49758689,"kids":[49763386,49760261,49763299,49759955,49760872,49758973,49759005,49761667,49759279],"score":60,"time":1789758006,"title":"The Implications of Linguistic Illegibility for LLM Security","type":"story","url":"https://arxiv.org/abs/2609.02852"},{"by":"Jimmc414","descendants":60,"id":49755533,"kids":[49760155,49755611,49755716,49759142,49759666,49755534,49759624,49759968,49758475,49756132,49761466,49761443,49759401,49759385],"score":109,"text":"https://archive.is/1CFX9
https://med.stanford.edu/news/all-news/2026/09/two-separate-...
Preprint of paper: https://www.biorxiv.org/content/10.1101/2025.07.02.662771v1","time":1789744327,"title":"Two parallel neural ectoderm progenitors contribute to the developing brain","type":"story","url":"https://www.newscientist.com/article/2589739-our-brain-evolved-from-two-primitive-nervous-systems-that-merged/"},{"by":"ibobev","descendants":216,"id":49746406,"kids":[49756261,49756157,49756256,49756241,49760384,49756290,49758369,49755527,49763133,49755594,49757357,49758312,49758680,49749248,49756113,49758664,49757815,49758119,49757522,49759351,49756343,49757383,49755615,49758960,49755632,49755584,49758729,49755869,49758698],"score":155,"time":1789678353,"title":"C++26: Trivial infinite loops are no longer undefined behaviour","type":"story","url":"https://www.sandordargo.com/blog/2026/09/16/cpp26-trivial-infinite-loops"},{"by":"anigbrowl","descendants":12,"id":49762960,"kids":[49763355,49763360,49763371,49763005,49762989,49763140,49763083,49763059,49763276],"score":12,"time":1789787791,"title":"Executive order on H-1B nonimmigrant visa administration","type":"story","url":"https://www.whitehouse.gov/presidential-actions/2026/09/enhancing-program-integrity-and-integrity-and-interagency-coordination-in-the-administration-of-the-h-1b-nonimmigrant-visa-program/"},{"by":"rbanffy","descendants":156,"id":49755160,"kids":[49757714,49755924,49756080,49755640,49762738,49758688,49756503,49757480,49756227,49756420,49757361,49756173,49755603,49756729,49757123,49756634,49758402,49756857],"score":193,"text":"https://archive.is/xjmvA","time":1789742732,"title":"North Korean nuclear test sets off years of earthquakes","type":"story","url":"https://www.science.org/content/article/north-korean-nuclear-test-sets-years-earthquakes"},{"by":"kingcauchy","descendants":17,"id":49714157,"kids":[49760372,49760229,49715368,49715355,49714272,49761317,49761295,49762947,49760126],"score":56,"time":1789486533,"title":"A search-and-inference database from scratch in pure Zig","type":"story","url":"https://antfly.io/research/antfly-zig"},{"by":"wittydeveloper","descendants":15,"id":49756671,"kids":[49756672,49762455,49757041,49757863,49756952,49757218,49757020,49756970],"score":68,"time":1789748723,"title":"We made Playwright 2x faster and 80% more token efficient","type":"story","url":"https://github.com/browserbase/stagehand"},{"by":"spacey","descendants":4,"id":49735057,"kids":[49761156],"score":25,"time":1789605949,"title":"Size-Specialized Memory Allocation","type":"story","url":"https://go.dev/blog/size-specialized-allocations"},{"by":"atarus","id":49753131,"score":1,"time":1789732810,"title":"Cekura (YC F24) Is Hiring","type":"job","url":"https://www.ycombinator.com/companies/cekura-ai/jobs/AiWwUxI-forward-deployed-engineer-us"},{"by":"m-hodges","descendants":195,"id":49755024,"kids":[49757458,49755587,49757275,49756033,49762957,49756044,49757677,49755234,49761916,49758019,49755957,49759551,49758004,49755858,49760356,49755625,49757737,49755388,49755397,49760029,49763138,49759059,49758058,49755855,49761771,49759398,49755906,49761835,49755841,49756255,49755415,49755664,49755356],"score":220,"text":"https://github.com/gaearon/conway-refinement#why-i-think-its...","time":1789742170,"title":"I vibed a proof of Conway's conjecture","type":"story","url":"https://overreacted.io/how-i-vibed-a-proof-of-conways-conjecture/"},{"by":"ibobev","descendants":19,"id":49755095,"kids":[49761784,49758678,49756806,49759774,49759489,49757401,49756849],"score":73,"text":"https://arxiv.org/abs/2510.20765","time":1789742462,"title":"Mathematicians Build Long-Awaited Graph Sandwich","type":"story","url":"https://www.quantamagazine.org/mathematicians-build-long-awaited-graph-sandwich-20260918/"},{"by":"realsarm","descendants":318,"id":49757520,"kids":[49758757,49758852,49758238,49759032,49758972,49759426,49758572,49758422,49758048,49759795,49759986,49758097,49759196,49757664,49757666,49762131,49759408,49758155,49759118,49758123,49761837,49759635,49758429,49761767,49759758,49758994,49762059,49762048,49761370,49758681,49758511,49758243,49758167,49760313,49759516,49758750,49758803,49759381,49759014,49759197,49758447,49759051,49759937,49762697,49758440],"score":424,"time":1789752481,"title":"US Military had close call after using AI for hallucinated intelligence report","type":"story","url":"https://www.cnn.com/2026/09/18/politics/us-military-ai-false-intelligence-china-ship"},{"by":"Matthalp","descendants":6,"id":49729631,"kids":[49733114,49762228,49729758],"score":23,"time":1789576957,"title":"Column built an issuer processor from scratch","type":"story","url":"https://column.com/card-issuing/"},{"by":"surprisetalk","descendants":0,"id":49740371,"score":30,"time":1789651203,"title":"From Geometry to Algebra and Back Again: 4000 Years of Papers (2023) [video]","type":"story","url":"https://www.youtube.com/watch?v=1cRFfYQYGxE"},{"by":"amirmc","descendants":2,"id":49759733,"kids":[49763300,49759979],"score":16,"time":1789762917,"title":"Human brain is two separate organs","type":"story","url":"https://med.stanford.edu/news/all-news/2026/09/two-separate-brains.html"},{"by":"Xyra","descendants":22,"id":49748041,"kids":[49759078,49759920,49760028,49757277,49761097,49762681,49758377,49757027,49758662,49759905,49759384,49758153,49759187],"score":48,"text":"Meet Scry, a 500 TB NVMe internet index in ClickHouse that you can run ~arbitrary readonly SQL and some of Datalog over, and I handle the problem of resource-contention with congestion-based micro-auction pricing. When there's capacity, the service is free for non-commercial use.
---
Hello. It's 2026, we're training simulated fruit fly brains to play Beat Saber, do we still have to be stuck with internet (re)search as fn: natural language -> black box we can't do anything about -> ranked_list/summary?
There is a long history of people trying to do very fancy things that end up being done in relational databases and a little SQL. There is a gravity to them, a bitter lesson, just like scaling of generalized ml training methods. I mean many, many information products can be built off essentially giant real-time OLAP databases and frontier LLMs writing brilliant SQL+Datalog+vector+Jev etc. queries.
Google Search, Tavily, Exa essentially have the problem of mapping your agents' context you are willing to provide, to a tiny subset of their index. You pay a fixed cost to an extremely hard problem that has a distribution of hardness, which means YOU eat the downsides when they are running out of budgeted compute to help you out.
Their algorithms are opaque to the caller, there's really not much user control, and there's not a serious opportunity to communally improve search recipes, like the lexical+Jev recipes you trust to select bleeding edge AI builders.
Furthermore, search companies aren't even pursuing text-to-SQL anymore (several have talked to me)... they made up their minds during the traumatic 2024 text-to-sql days. They were just too early.
I hope you enjoy. I'm intent on scaling this paradigm on differentiated hardware over much more data, so any compelling use cases or queries I could show off, would be much appreciated!","time":1789686957,"title":"Show HN: Scry, programmable internet search w/ congestion pricing","type":"story","url":"https://scry.io/"},{"by":"rbanffy","descendants":97,"id":49754250,"kids":[49755084,49754816,49754888,49754911,49755383,49754781,49755091,49759090,49755574,49755175,49761267,49755341,49759785,49754784,49754889,49754756,49758424,49754822,49754753,49755192,49755113,49754859,49754862,49755966,49758866,49754787],"score":93,"time":1789738767,"title":"AI chatbots are becoming experts at changing people's minds","type":"story","url":"https://www.science.org/content/article/ai-chatbots-are-becoming-experts-changing-people-s-minds-what-s-their-secret"},{"by":"ethanhawksley","descendants":725,"id":49753211,"kids":[49753634,49753722,49758796,49754918,49753481,49754147,49754048,49753552,49753532,49753463,49755765,49755242,49755986,49754579,49755427,49753738,49753529,49753551,49757874,49754343,49757453,49755032,49753466,49754290,49758099,49753984,49754732,49753766,49759790,49757076,49762134,49754013,49754425,49759149,49760424,49759452,49756616,49753539,49754265,49756498,49759742,49758860,49754171,49753685,49759505,49753504,49754996,49753640,49753426,49754401,49755244,49757911,49753674,49754077,49758036,49757414,49756180,49754597,49756437,49757327,49754794,49755295,49758284,49757058,49757672,49754274,49757017,49755604,49756468,49757733,49755622,49756572,49755169,49758439,49756181,49754855,49755315,49755284,49753849,49755686,49754299,49754056,49756949,49754790,49761169,49759523,49755272,49753688,49760263,49753622,49753541,49754236,49756222,49754109,49759857,49753464,49755505,49757749,49754375,49753891,49754042,49754181,49754718,49754344,49753672,49757494,49754934,49755562,49755048,49759587,49758624,49753734,49753459,49754087,49756309,49755860,49755455,49755023,49753936,49754398,49754848,49755273,49754572,49753928,49753748,49754883,49756724,49754613,49757247,49755489,49754439,49754208,49762122,49753908,49754391,49753570,49754341,49754310,49754556,49754761,49753561,49754870,49755803],"score":745,"time":1789733210,"title":"I don't like passkeys","type":"story","url":"https://hawksley.dev/blog/i-dont-like-passkeys"},{"by":"vquemener","descendants":94,"id":49727245,"kids":[49758310,49754282,49756071,49760645,49753271,49753499,49756347,49754778,49762603,49752711,49752736,49761602,49752628,49752717,49759965,49756716,49752122,49753869,49754069,49754585,49752611,49755956,49752734,49752966,49753678,49752359,49728286,49752418,49754956,49754484,49753690,49752891,49753637,49754583,49754724,49752974],"score":135,"text":"https://delta.dev/","time":1789567544,"title":"Replacing Pull Requests with Delta","type":"story","url":"https://zed.dev/blog/delta-public-beta"},{"by":"arnemunthekaas","descendants":255,"id":49711544,"kids":[49717700,49715349,49713097,49714884,49739929,49715696,49718953,49713446,49723421,49714603,49712016,49718691,49722697,49713017,49719062,49713041,49712181,49715201,49733119,49725884,49711987,49745977,49718181,49713020,49712366,49718605,49712641,49721031,49737285,49744101,49727042,49714844,49718930,49725759,49720512,49718923,49713187,49715275,49718108,49714776,49714283,49712421,49716106,49723550,49725175,49723223,49746990,49713437,49730213,49714858,49737153,49722586,49723977,49715159,49713293,49718654,49743541,49712498,49712762,49727866,49724266,49720971,49713240,49724382,49712292,49718837,49719799,49714300,49712734,49715327,49723293,49721826,49719948,49712937,49729225,49726356,49720006,49732665,49715760,49712230,49718147,49713465,49720620,49724389,49719021,49714737,49714370,49733570,49712007,49723237,49715033,49712788,49723815,49711546,49712388,49756854,49740197,49731465,49726559,49721245,49722999,49754944,49721418,49712290,49733500,49730991,49722997,49722993,49713595,49723362,49716038,49728296,49714819,49712999,49713035,49714325,49712803,49713891,49717053],"score":2322,"text":"See also: Avian Visitors - https://news.ycombinator.com/item?id=48343424 - May 2026 (20 comments)","time":1789475470,"title":"Show HN: An e-ink frame that hears birds and draws them as 1800s illustrations","type":"story","url":"https://github.com/arnegiacomo/fugleramme"},{"by":"axiologist","descendants":67,"id":49749771,"kids":[49750430,49751755,49750463,49751826,49751962,49751308,49753520,49750392,49751092,49752696,49750395,49751198,49751595,49753171,49750406],"score":174,"time":1789700678,"title":"Pre-Greek: The lost language hidden within Ancient Greek","type":"story","url":"https://linguisticdiscovery.com/posts/pre-greek/"},{"by":"saimiam","descendants":194,"id":49752614,"kids":[49753930,49753047,49753094,49753152,49757265,49753350,49754336,49754519,49756686,49756768,49752926,49758847,49760078,49752826,49756367,49758875,49758159,49757022,49753820,49760065,49759885,49755659,49754715,49755848,49760020,49753424,49753002,49753664],"score":293,"time":1789729304,"title":"Warren Buffett Steps Down as Berkshire Chairman, Names Son to Replace Him","type":"story","url":"https://www.nytimes.com/2026/09/18/business/warren-buffett-berkshire-chairman.html"},{"by":"0x63_Problems","descendants":34,"id":49744416,"kids":[49760769,49759235,49758451,49759700,49758478,49758495,49758411,49759040,49758472,49758479],"score":29,"text":"I'm the co-founder of Gauge, and I built ax-check.com to quickly test how well coding agents can onboard to your product.
You'll get a scorecard, specific suggested fixes, and three full coding sessions that show how agents read your site and use your product.
I built this because similar checks were too noisy. Most suggested obscure technical changes that don't actually make a difference in agent experience (or AX, hence ax-check.com).
This check starts by using DeepSeek 4.1 Flash to try to find key information about your product, starting from the homepage. In actual agent traffic data, we've seen that the key pages are the homepage, llms.txt, pricing, and the docs site (by traffic volume, and by influence), so we focus on those and ignore the rest. We also find that content negotiation for Markdown is legitimately helpful for agents to complete tasks faster and find what they're looking for, so the scan tests that your key pages can serve Markdown.
The other key piece is that we run actual coding agents in sandboxes, and have them try to onboard to your product. You can see the full trace and watch it happen live (we kick it off fresh when you enter a new site). We surface interesting findings like hallucinated URLs, inaccurate docs instructions, or product confusion.
It also detects whether the agents could complete a fully working onboarding autonomously, without being blocked by a login wall. This is still controversial, but I think finding ways to let agents safely onboard autonomously is going to be table stakes within a year for developer tools in particular.
The whole site is agent-friendly itself! You can generally just talk to your coding agent about ax-check.com and it can do the rest. Would really appreciate any feedback to make this useful.","time":1789668482,"title":"Show HN: Ax-check.com – Can agents use your product?","type":"story","url":"https://www.ax-check.com/"},{"by":"saikatsg","descendants":2,"id":49757883,"kids":[49759950],"score":22,"time":1789754179,"title":"Ctenophores Aren't Just Beautiful. They're Biological Wonders","type":"story","url":"https://www.quantamagazine.org/ctenophores-arent-just-beautiful-theyre-biological-wonders-20260916/"},{"by":"throw7","descendants":90,"id":49759466,"kids":[49759886,49759860,49759904,49761216,49761780,49761964,49760918,49762419,49760963,49761520,49760134,49760493,49759879,49759807,49761075,49760158,49760360,49759953],"score":294,"time":1789761774,"title":"Korea raises data breach fines to 10% of revenue","type":"story","url":"https://www.koreajoongangdaily.com/business/korea-raises-data-breach-fines-to-10-of-revenue/12869899"},{"by":"Handy-Man","descendants":198,"id":49749656,"kids":[49750434,49750433,49750602,49750975,49751173,49750117,49750498,49750873,49749989,49749962,49754582,49750355,49750474,49749996,49751073,49759340,49750091,49750780,49750155,49751918,49751366,49751355,49751811,49751226,49750631,49751055,49750153,49750412,49758130,49751230,49750981,49749954,49751372,49751258,49749950,49750517,49749660,49751282,49751897,49749976,49750727,49753971,49750119,49751333],"score":474,"time":1789699644,"title":"A heap overflow and SSO misconfiguration to compromise OpenAI internal repos","type":"story","url":"https://www.hacktron.ai/blog/hacking-openai"},{"by":"ledoge","descendants":3,"id":49761866,"kids":[49763017],"score":25,"time":1789775943,"title":"Prepare your iPhone 18 Pro Max to ship","type":"story","url":"https://support.apple.com/en-us/127848"},{"by":"vismit2000","descendants":13,"id":49721507,"kids":[49754884,49755832,49757004,49758360,49754316,49754606],"score":63,"time":1789526242,"title":"The Shadows Lurking in the Equations – Underwater Islands","type":"story","url":"https://gods.art/articles/equation_shadows"},{"by":"asplake","descendants":24,"id":49754064,"kids":[49755577,49755282,49755943,49755373,49757010,49756221,49762916],"score":51,"time":1789737875,"title":"NATS publishes preliminary report on technical incident of 8 September","type":"story","url":"https://www.nats.aero/news/nats-publishes-preliminary-report-on-technical-incident-of-8-september/"},{"by":"vinhnx","descendants":18,"id":49726614,"kids":[49751807,49753111,49752750,49752897,49752931,49753178],"score":104,"time":1789564815,"title":"When the fractional part of a float fixes your shader","type":"story","url":"https://crocidb.com/post/when-the-fractional-part-of-a-float-fixes-your-shader/"},{"by":"hentrep","descendants":12,"id":49761654,"kids":[49763030,49762711,49761992,49762855],"score":21,"time":1789774250,"title":"Uber ordered to pay $40M over death of woman left on Southern California freeway","type":"story","url":"https://ktla.com/news/california/uber-driver-abandoned-rider-freeway/"},{"by":"JonSchneider","descendants":189,"id":49746618,"kids":[49747390,49747481,49751434,49747164,49747732,49751923,49747684,49759096,49749126,49748190,49751118,49746767,49748869,49754146,49750047,49747356,49753948,49754994,49749483,49749328,49748450,49747586,49748117,49753470,49748346,49752219,49748463,49747221,49749801,49748752,49746790,49749837,49753562,49747858,49750162,49752181,49751741,49751747,49751591,49750453,49754204,49752180,49749583,49748445,49749822,49748586,49750937],"score":565,"time":1789679611,"title":"Bonsai 2 27B: Near-Lossless Compression in a 9x Smaller Footprint","type":"story","url":"https://prismml.com/news/bonsai-2-27b"},{"by":"embedding-shape","descendants":60,"id":49710447,"kids":[49749104,49743181,49710453,49752443,49749791,49752096,49748887,49749316,49748435,49750593,49752706,49750629,49747781],"score":98,"time":1789469184,"title":"Diplodocus, Long Thought Exclusively American, Turns Up in Spain","type":"story","url":"https://www.sci.news/paleontology/spanish-diplodocus-15064.html"},{"by":"vertigoruntime","descendants":671,"id":49745940,"kids":[49748312,49748503,49747136,49746363,49746353,49746093,49749153,49747322,49748571,49748280,49749892,49746376,49746213,49746794,49748162,49746869,49750106,49746703,49750132,49748480,49746704,49749676,49758304,49746556,49749386,49746031,49747501,49750431,49751959,49749412,49750415,49748073,49749433,49762499,49752167,49747803,49749625,49746079,49752407,49748137,49753427,49747523,49750242,49747515,49753189,49748129,49748967,49746115,49747273,49748555,49747291,49746709,49751882,49749964,49749073,49747693,49749142,49747299,49749529,49752073,49753419,49746028,49755186,49748746,49751672,49748491,49754477,49747520,49746834,49751119,49752623,49747196,49748709,49748437,49749052,49749796,49749399,49746244,49746498,49747023,49747165,49746477,49748788,49747278,49746090,49750674,49747599,49746392,49748143,49746104,49748911,49747333,49748585,49749441,49750101,49748679,49746464,49746927],"score":567,"time":1789676268,"title":"Astra for Law","type":"story","url":"https://openai.com/index/astra-for-law/"},{"by":"drgo","descendants":15,"id":49751222,"kids":[49762106,49755235,49755176,49755300,49756307,49758124,49762569,49753784],"score":60,"time":1789716770,"title":"New wild cat species discovered – with only one known living member","type":"story","url":"https://www.bbc.com/news/articles/c6x2zgv9rr4ro"},{"by":"usernomdeguerre","descendants":48,"id":49762493,"kids":[49762856,49762914,49763169,49762854,49762595,49762999,49763175,49762761,49762806,49762978,49763104,49763237,49763130,49763113,49763198,49763272,49762785,49762587,49763330,49762971,49762893,49763285],"score":40,"time":1789782035,"title":"Gemini hacked three companies in first known breakout by Google's AI","type":"story","url":"https://www.reuters.com/business/gemini-hacked-three-companies-first-known-breakout-by-google-ai-wsj-reports-2026-09-18/"},{"by":"imalessandro","descendants":1,"id":49757757,"kids":[49761071],"score":17,"time":1789753581,"title":"Show HN: Jeff – A read-only CLI for semantic code review using Jev","type":"story","url":"https://github.com/Alurith/jeff"},{"by":"mmh0000","descendants":168,"id":49758028,"kids":[49758733,49759504,49758264,49758685,49760471,49758759,49758568,49758323,49759566,49759124,49758891,49761814,49762344,49760115,49761073,49759045,49761227,49758255],"score":195,"time":1789754912,"title":"Border agents can search cellphones without a warrant or reasonable suspicion","type":"story","url":"https://lawandcrime.com/high-profile/the-government-was-entitled-trumps-border-agents-can-now-search-cellphones-without-a-warrant-probable-cause-or-reasonable-suspicion-2nd-circuit-rules/"},{"by":"gkfasdfasdf","descendants":93,"id":49750152,"kids":[49750698,49755709,49751284,49752019,49752654,49758466,49751655,49756547,49751456,49750627],"score":325,"time":1789705224,"title":"Jemalloc 5.4.0","type":"story","url":"https://github.com/jemalloc/jemalloc/releases/tag/5.4.0"},{"by":"absqueued","descendants":80,"id":49746290,"kids":[49750765,49751260,49752184,49749217,49748180,49749544,49747983,49748640,49763213,49750521,49746570,49750819,49749101,49748839,49748409,49752857,49747375,49753382,49750015,49749546,49748229],"score":160,"time":1789677868,"title":"Flet 1.0 – Build cross-platform apps in Python","type":"story","url":"https://flet.dev/"},{"by":"bookofjoe","descendants":192,"id":49743097,"kids":[49744288,49744555,49744185,49761173,49744877,49747417,49744095,49752010,49743424,49745537,49745146,49745220,49744375,49744590,49755557,49744162,49744538,49744870,49750953,49750398,49744260,49752366,49752144,49751861,49758899,49753383,49748621,49748078,49747744,49750232,49749107,49750407,49745849,49744406,49749440,49750300,49745605,49747576,49744194,49748961,49750636,49744906,49743561,49745322,49754640,49744680,49748848,49745024,49743596,49745080,49751603,49745011,49746224,49745250,49754221,49750688,49753966,49751866,49745690],"score":698,"time":1789662337,"title":"Hister: A private search engine for the pages you visit and the files you keep","type":"story","url":"https://github.com/asciimoo/hister"},{"by":"LiamPowell","descendants":229,"id":49753179,"kids":[49754976,49753379,49753638,49754665,49754974,49753798,49753462,49753364,49753407,49755111,49753432,49757991,49755114,49753898,49753498,49755805,49754062,49756418,49753461,49753864,49756212,49753428,49754288,49754728,49756424,49753472,49753372,49753852,49753469,49753414,49754815,49756720,49753809,49755556,49754476,49754650,49753914,49754132,49753761,49754566,49754866,49753785,49754433,49753381,49758220,49753624,49754001,49754760,49754764,49755252,49755119,49753963,49754122,49756188,49753804,49755018],"score":313,"time":1789733035,"title":"Bend 2 and the Vibe-Coding Trap","type":"story","url":"https://blog.liampwll.com/posts/bend_vibe_coding/"},{"by":"ibobev","descendants":9,"id":49758709,"kids":[49760812,49759599,49762672,49762273,49759495,49758846],"score":54,"time":1789758063,"title":"How did AMD Ryzen get 50% faster in two years?","type":"story","url":"https://lemire.me/blog/2026/09/18/how-did-amd-ryzen-get-50-faster-in-two-years/"},{"by":"my123","descendants":246,"id":49715813,"kids":[49742389,49743357,49742842,49740724,49744607,49740600,49751328,49745907,49740729,49743850,49741593,49741793,49741859,49742977,49749890,49741367,49749318,49740870,49743953,49740966,49749442,49744951,49743022,49742530,49741888,49741212,49740621,49743884,49748018,49745367,49744325,49741145,49748164,49742577,49741089,49742408],"score":626,"time":1789493318,"title":"Fujitsu launches made-in-Japan next-generation CPU FUJITSU-MONAKA","type":"story","url":"https://global.fujitsu/en-global/pr/news/2026/09/14-02"},{"by":"dagmx","descendants":88,"id":49750094,"kids":[49751784,49750760,49760778,49751314,49751138,49751886,49758329,49755861,49754635,49754303],"score":279,"time":1789704588,"title":"The scourge of x86 emulation","type":"story","url":"https://fex-emu.com/Scourge-of-emulation/"},{"by":"wek","descendants":57,"id":49753878,"kids":[49754447,49755993,49754869,49757146,49755940,49760342,49760258,49754507,49756349,49754688,49757555,49755969,49762199,49759164,49759218,49756734,49755865,49755418,49756916,49754167,49757620],"score":204,"time":1789736790,"title":"An empirical study of harness design for coding agents","type":"story","url":"https://arxiv.org/abs/2609.20804"},{"by":"mhb","descendants":106,"id":49726007,"kids":[49746411,49746119,49745835,49745590,49749291,49746336,49747003,49752587,49745854,49749489,49746712,49749455,49748658,49749727,49747670,49754377,49745739,49748365,49746398,49752012,49745878,49746941,49749287],"score":504,"time":1789562114,"title":"Wax motor","type":"story","url":"https://en.wikipedia.org/wiki/Wax_motor"},{"by":"squirrel","descendants":3,"id":49760971,"kids":[49762950,49763357,49762408],"score":22,"time":1789769735,"title":"US and Denmark reach deal over Greenland security","type":"story","url":"https://www.bbc.co.uk/news/articles/c63d7lexyym1o"},{"by":"jimjag","descendants":18,"id":49754823,"kids":[49757269,49756116,49758494,49758713,49756370,49758088,49754824,49755712,49755722,49755947],"score":34,"time":1789741284,"title":"BeanShell3 in Development","type":"story","url":"https://beanshell.github.io/home.html"},{"by":"ChrisArchitect","descendants":51,"id":49746407,"kids":[49751158,49748685,49752245,49748124,49751246,49746933,49760669,49753546,49749377,49751115,49749469,49750766,49751383,49748418,49756031,49748701,49746935,49751094,49746485,49750265,49749458,49748431,49748200,49751830,49756668,49749963,49750495],"score":146,"time":1789678356,"title":"The most important product decision is what you don't build","type":"story","url":"https://liamnugent.me/posts/what-you-dont-build/"},{"by":"pseudolus","descendants":442,"id":49740260,"kids":[49741347,49745790,49740828,49741300,49742268,49745762,49744033,49746042,49741525,49743029,49744736,49751439,49745950,49741250,49746125,49754413,49757785,49744447,49742992,49741068,49745209,49751815,49755212,49745510,49746199,49744650,49748761,49749826,49741118,49747703,49741282,49745686,49741330,49741376,49741176,49740872,49744284,49748884,49746522,49744667,49745281,49744390,49740804,49740668,49741280,49756721,49740261,49744107,49742871],"score":257,"text":"https://archive.ph/F77tj","time":1789650748,"title":"The American Religion of Self-Storage Facilities","type":"story","url":"https://www.newyorker.com/magazine/2026/09/21/the-american-religion-of-self-storage-facilities"},{"by":"pluc","descendants":769,"id":49752056,"kids":[49753540,49752642,49752405,49752551,49753032,49753746,49753737,49752941,49758587,49756015,49752444,49753673,49755476,49753162,49752172,49756972,49752656,49756678,49752412,49757600,49756679,49752769,49762030,49753021,49754499,49753333,49753320,49757889,49752990,49753843,49752786,49761887,49759413,49752242,49753996,49752310,49762613,49757210,49753331,49752603,49752375,49752386,49752718,49753220,49757370,49753044,49752978,49752406,49754218,49760398,49753921,49755792,49753581,49752695,49755926,49758354,49755481,49755025,49754563,49752859,49753011,49752233,49758574,49754590,49754088,49754581,49756659,49758581,49754279,49753989,49753387,49753856,49754954,49758443,49753205,49752368,49753010,49756110,49752228,49753374,49752309,49752604,49752670,49753149,49755118,49752735,49752624,49753655,49753129,49756317],"score":871,"time":1789724707,"title":"Microsoft exec called AI scraping 'the largest theft of labor in human history'","type":"story","url":"https://techcrunch.com/2026/09/17/microsoft-exec-called-ai-scraping-the-largest-theft-of-labor-in-human-history-new-unredacted-filings-reveal/"},{"by":"jjcm","descendants":128,"id":49747925,"kids":[49751471,49749728,49749361,49749246,49750958,49758075,49751910,49749162,49749924,49750537],"score":328,"time":1789686348,"title":"Qwen 3.8 Omni Flash","type":"story","url":"https://qwen.ai/blog?id=qwen3.8-omni-flash"},{"by":"aidog","descendants":5,"id":49755431,"kids":[49757931,49756818,49758575,49756810,49757002,49757506],"score":28,"time":1789743894,"title":"GrassLobster: AI Agentic Generation of Parametric Geometry Workflows","type":"story","url":"https://www.miro.vision/index.php/2026/09/17/grasslobbster/"},{"by":"HotGarbage","descendants":72,"id":49754687,"kids":[49755916,49755878,49755713,49755499,49755500,49755220,49755149,49758207,49757767,49755350,49756154,49756382],"score":118,"time":1789740682,"title":"Second Circuit allows government to search electronic devices at the border","type":"story","url":"https://knightcolumbia.org/content/second-circuit-allows-government-to-search-electronic-devices-at-the-border-without-any-suspicion"},{"by":"TMWNN","descendants":32,"id":49748957,"kids":[49750928,49749818,49749526,49751422,49749366,49756381,49749847,49750851,49750400,49749854,49749360,49749895],"score":89,"time":1789693503,"title":"Telstra outage: The night a network decided the year was 2006","type":"story","url":"https://www.netnod.se/blog/telstra-outage-night-network-decided-year-was-2006"},{"by":"robotnikman","descendants":3,"id":49760758,"kids":[49762677,49762558],"score":26,"time":1789768305,"title":"Valve has open-sourced Lepton, its tool to bring Android games to Steam","type":"story","url":"https://www.theverge.com/tech/997315/valve-has-open-sourced-lepton-its-tool-to-bring-android-games-to-steam"},{"by":"syntaxing","descendants":38,"id":49749393,"kids":[49750439,49753203,49752133,49754145,49750491,49752177],"score":97,"time":1789697059,"title":"Shapelearn Qwen 3.8 27B (13.1 GB VRAM)","type":"story","url":"https://byteshape.com/blogs/Qwen3.8-27B/"},{"by":"darkwater","descendants":127,"id":49742353,"kids":[49743432,49745127,49742964,49743281,49742732,49742686,49743706,49743122,49744509,49742762,49743719,49743098,49744893,49745609,49750473,49751669,49743765,49743201,49743194,49747166,49743518,49744868,49744103,49743308,49742748,49748866,49742758,49747722],"score":172,"time":1789659234,"title":"Rate limits on GitLab.com are changing","type":"story","url":"https://about.gitlab.com/blog/rate-limit-change-2026/"},{"by":"nonmaskable","descendants":402,"id":49724881,"kids":[49734146,49739057,49733875,49735409,49738821,49734204,49740396,49761055,49736787,49735862,49735291,49737170,49748299,49733954,49736116,49738446,49734101,49735842,49733856,49745679,49733918,49737786,49734075,49736597,49734454,49735884,49743779,49735114,49738564,49739993,49734769,49737092,49738013,49734484,49737352,49742565,49724907,49738774,49734579,49725272,49748637,49734474,49734980,49735390],"score":958,"time":1789557353,"title":"Nvidia announces native GPU programming in Rust","type":"story","url":"https://developer.nvidia.com/blog/introducing-cuda-rust-two-tracks-for-writing-gpu-kernels/"},{"by":"Capira","descendants":45,"id":49746346,"kids":[49747639,49749043,49750515,49750104,49748086,49747801,49752211,49749998,49751301,49747340,49747764,49753411,49750425,49747784,49748827,49749099,49748851,49749276,49748736,49748616,49756609,49750293],"score":97,"time":1789678084,"title":"Show HN: Snapdrop: Instantly share files between devices. No setup, no signup","type":"story","url":"https://snapdrop.me"},{"by":"13613288957","descendants":31,"id":49749171,"kids":[49763238,49751321,49749909,49750334,49750315,49750522,49749851,49750710,49752784,49750948,49750732,49750711,49749901,49751872,49749172,49750267,49753962,49752191,49749974,49752913,49750116,49750146],"score":58,"time":1789695248,"title":"Ask A Monk – A digital wilderness for thoughts with no immediate answer","type":"story","url":"https://askamonk.online"},{"by":"nadis","descendants":3,"id":49755337,"kids":[49763197,49757435],"score":25,"time":1789743489,"title":"Build Faster Feedback Loops Using Qualitative User Research","type":"story","url":"https://blog.nseldeib.com/p/build-faster-feedback-loops-using"},{"by":"lr0","descendants":6,"id":49760749,"kids":[49762328,49762293,49761555,49762182,49762207,49760821],"score":20,"time":1789768252,"title":"I Hate Workday","type":"story","url":"https://nelson.cloud/i-hate-workday/"},{"by":"nicolas-siplis","descendants":303,"id":49746163,"kids":[49746605,49749199,49749090,49752592,49749040,49747163,49752912,49746985,49754515,49746554,49746461,49747521,49746915,49747710,49754838,49754473,49747045,49747459,49755626,49747113,49746784,49752965,49749261,49747145,49750799,49749520,49746889,49750014,49746717,49751744,49750599,49748456,49749948,49750811,49752036,49755979,49748536,49749002,49752447,49748755,49754240,49749794,49754586,49746930,49748526,49747760,49755812,49746592,49747389,49755021,49752070,49746947,49747664,49749902,49746480,49754559,49746888,49751201,49747092,49747940,49750618,49747408,49746839,49747068,49752082,49746824,49746921,49757368,49749214,49752307,49747542,49747084,49752523,49747161,49748820,49749000,49747502,49746515,49746447,49748091,49750750,49755038,49749553,49747464,49748921,49752311,49748373,49756472,49748590,49756887,49754734,49749182,49750060,49746997],"score":593,"time":1789677373,"title":"Bend – a language that blocks AI mistakes via proof and runs on GPUs","type":"story","url":"https://bend-lang.com/"},{"by":"steveybrown","descendants":135,"id":49740105,"kids":[49741747,49744177,49744102,49761819,49742832,49742652,49749378,49741664,49749222,49744479,49743520,49749702,49742154,49742220,49742775,49742442,49748696,49742440,49744142,49742906,49745673,49749050,49744679,49749979,49742178,49744318,49741654,49748349,49744377,49741896,49745939,49742130,49741883,49743954,49743672,49748364,49748016,49745389,49742629,49749075,49743757,49746440,49744350,49743621,49746001,49746638,49742946,49744911,49743134,49744077,49743941,49750229,49744302],"score":231,"text":"I kept seeing engineers share what they were building with AI; however, I was always more curious about how they worked. Which agents did they use? What skills and tools had stuck or been thrown out the window? How did they manage longer-running tasks? So I built this with the hope we could have a dedicated space to share and be open about our setups.","time":1789650104,"title":"Show HN: Share your AI Setup, Learn from others","type":"story","url":"https://mysetup.ai/"},{"by":"vinhnx","descendants":6,"id":49762617,"kids":[49762804],"score":9,"time":1789783734,"title":"Elon Musk and Tesla Forged a New EV Path","type":"story","url":"https://spectrum.ieee.org/elon-musk-tesla"},{"by":"iscmt","descendants":49,"id":49746628,"kids":[49753515,49748949,49749017,49747895,49747621,49751498,49750477,49751881,49758731,49749201,49748802,49751825,49750142],"score":112,"time":1789679672,"title":"How Uber Protects Against Retry Storms","type":"story","url":"https://www.uber.com/us/en/blog/protecting-against-retry-storms/"},{"by":"osnium123","descendants":53,"id":49714096,"kids":[49749484,49722820,49747349,49716847,49745604,49722162,49745597,49745279,49714097,49717881,49716673],"score":120,"text":"Abstract at IEDM website below
ABSTRACT
A14 NanoFlexTM Pro platform technology features the world’s smallest SRAM with a cell size <0.017μm2 and highest SRAM macro density to date through dimensional ground rule scaling. Coupled with standard cell innovations, it delivers a full-node PPA improvement over the predecessor N2 technology for a 10~15% speed enhancement, a 25~30% power reduction and a chip density increase by ~20% driven by logic and SRAM scaling. Key features, such as TSV, innovative RDL that enables SoIC bond pitch of 4.5μm and high-performance MiM processes, have also been developed to accelerate the system-level integration and scaling for SoIC products. Supported by robust yield and device progress, A14 technology is on track for volume production in 2028.","time":1789486315,"title":"TSMC revealing details about next gen A14 node","type":"story","url":"https://iedm26.mapyourshow.com/8_0/sessions/session-details.cfm?scheduleid=331"},{"by":"vertigoruntime","descendants":2,"id":49761392,"kids":[49763382,49762514],"score":11,"time":1789772395,"title":"OpenAI expects to burn through almost $280B by 2030, FT reports","type":"story","url":"https://www.reuters.com/technology/openai-expects-burn-through-almost-280-billion-by-2030-ft-reports-2026-09-18/"},{"by":"cat-whisperer","descendants":54,"id":49743049,"kids":[49750056,49744434,49762052,49746149,49744699,49749420,49745666,49747758,49743693,49745896,49745338,49744147,49745346,49745147,49744429,49749240,49745875,49751689,49745971],"score":63,"text":"Hey HN, we're Nars & Nishant, founders of Skillsync (https://skillsync.com)
Skillsync lets you move your AI chats across every coding agent. Most of our work exists as conversations, which are currently scattered across our agents. Though stored locally, these conversations use different formats. This is annoying because you cannot simply switch between agents without starting over. We get locked into a single provider and their agent as we invest in skills and memories over time. Skillsync acts as a universal converter. It moves the entire session, including all the messages, reasoning and tool calls so you can pick up right where you left off.
Skillsync collects all your sessions in one place and makes them searchable. It breaks down what each session is carrying, including which loaded skills the agent is actually using, making stale context easy to spot. You can also create shared workspaces to sync sessions across your team. You can build your own closed loop systems. Everything runs locally except when you share to workspaces.
The core is an open-source Rust engine called txcript (https://github.com/skillsynchq/txcript). It translates a session from one agent's on-disk format into another's, mapping conversation, reasoning, and tool history. Think ffmpeg or pandoc, but for agent sessions.
On top of that engine is a local-first desktop app. Your agent sessions are normally scattered across different tools' folders in formats you'd never read by hand; the app surfaces them in one place with a UI that makes them actually readable, the conversation, the reasoning, and the tool calls, so you can revisit what happened, move a session into another agent, or share it with a teammate.
Skills and memory are stored as portable, human-readable markdown you own, and exposed to any agent over MCP for search and selective retrieval.
Sessions and translation run locally on your machine. The one thing that leaves is what you explicitly share into a team workspace.
It’s been surprising to us to see how locked-in people already are without realizing it. You don't notice you're trapped in one agent or harness until you try to leave, and by then you've got months of sessions stuck in a format only that tool can read. The lock-in is invisible right up until it's expensive.
Once sessions are portable, they stop being disposable logs and become something you can actually build on, spotting patterns in how you work, handing a session off to someone else, letting a non-technical teammate pick up where an engineer left off. The session turns out to be the unit of collaboration, and right now it's being thrown away.
Before Skillsync, Nars and I built an open-source payment orchestrator that let merchants route across many processors instead of getting locked into one (30k+ GitHub stars). Fighting vendor lock-in by making incompatible systems interoperate was the whole job. We came into YC with a different idea, but the more we lived inside coding agents the more we saw the same problem from the other side: your context is locked into whatever agent you start with, because every agent stores sessions differently and none of it moves.
Some of the things we’re seeing people do with Skillsync are:
- Moving a session between Claude Code, Codex, and Cursor mid-task, including when you hit a usage limit on one and want to keep going on another.
- Thinking through a problem in a browser Claude chat, then handing the whole thread to Claude Code or Codex to build.
- Sharing research sessions with a team. Everyone uploads their sessions, so teammates and their agents can build on each other's work instead of repeating it.
It's available as a Mac app, CLI and MCP. Here's the demo: https://youtu.be/7hVhSnSKGl8. Would love to hear your feedback and answer questions!","time":1789662135,"title":"Launch HN: Skillsync (YC W26) – AI chat sessions made portable across agents","type":"story"},{"by":"idatum","descendants":3,"id":49761694,"kids":[49762320],"score":8,"time":1789774524,"title":"Yongsan Electronics Market (Seoul)","type":"story","url":"https://en.wikipedia.org/wiki/Yongsan_Electronics_Market"}]