15:[[48253321,{"by":"manitow","dead":true,"id":48253321,"kids":[48253391],"score":1,"time":1779585209,"type":"story"}],[48234090,{"by":"jetter","descendants":161,"id":48234090,"kids":[48234672,48234296,48238207,48236044,48239795,48235722,48234322,48234440,48241519,48238634,48238250,48234597,48241111,48238064,48244860,48237554,48239684,48245084,48246923,48234521,48235846,48234707,48242221,48237832,48237197,48234782,48234647,48238017,48247964,48235799,48234377,48245707,48236496,48235971,48234738,48241931,48235965,48264056,48235398,48235394,48237116,48256259,48236009,48234753,48234541,48237441],"score":421,"time":1779446306,"title":"Antigravity 2.0 Tops the OpenSCAD Architectural 3D LLM Benchmark","type":"story","url":"https://modelrift.com/blog/openscad-llm-benchmark/"}],[48219918,{"by":"shideneyu","descendants":97,"id":48219918,"kids":[48221033,48220069,48220403,48220156,48220954,48221444,48220228,48220211,48225822,48220931,48220929,48221106,48220111,48232736,48223268,48220864,48227105,48244555,48221237,48220330,48271727,48222102,48219925,48227660,48220981,48220290,48223663,48225623,48227805,48245886],"score":194,"text":"Author here. RMUX started from a frustration: I've used tmux for years and got tired of scraping output with grep and sleeps to automate anything. So I rebuilt the multiplexer from scratch in Rust, with a programmable layer on top.

Two surfaces: a tmux-compatible CLI (~90 commands, your keybindings just work), and a typed async Rust SDK on the same daemon — stable pane IDs, structured snapshots, locator-style waits. The idea is Playwright-style automation, but for terminals.

Native on Linux, macOS, Windows (real ConPTY, no WSL).

Demos and docs at rmux.io. Happy to answer questions about the daemon protocol, ConPTY, or the SDK design.","time":1779355320,"title":"Show HN: Rmux – A programmable terminal multiplexer with a Playwright-style SDK","type":"story","url":"https://github.com/helvesec/rmux"}],[48205626,{"by":"kevinsimper","descendants":290,"id":48205626,"kids":[48207654,48209290,48206924,48205897,48206573,48208295,48209844,48210341,48213542,48208809,48206631,48213054,48206452,48206963,48215372,48210181,48216318,48214995,48207416,48210616,48210064,48210002,48216712,48216959,48380630,48208123,48208653,48213047,48207195,48222367,48209977,48223635,48211321,48205917,48214095,48207005,48209117,48206954,48206952],"score":721,"time":1779273302,"title":"Qwen3.7-Max: The Agent Frontier","type":"story","url":"https://qwen.ai/blog?id=qwen3.7"}],[48169874,{"by":"Bibabomas","descendants":151,"id":48169874,"kids":[48175406,48175368,48172599,48174550,48175025,48177252,48174924,48173163,48173657,48178226,48205911,48173813,48173704,48173521,48175201,48173506,48182190,48176126,48176348,48175502,48190303,48177913,48176396,48181151,48176916,48170972,48171856,48173270,48173726,48174555,48175522,48177905,48176075,48176235,48175878,48176203,48175739,48176621,48173477,48181447,48176929,48175065,48180998,48178864,48171482,48174827,48177986,48179113,48175026,48176661,48180738,48180184,48180783,48180522,48183989,48179821,48193415,48194505,48176367,48177231,48178270,48178603,48176477,48179331,48190795,48175233,48179372,48176875,48175537,48185245,48178113,48179504,48176209,48173950,48173229],"score":445,"text":"$16","time":1779032227,"title":"Show HN: Semble – Code search for agents that uses 98% fewer tokens than grep","type":"story","url":"https://github.com/MinishLab/semble"}],[48130374,{"by":"andersonmvd","descendants":12,"id":48130374,"kids":[48131374,48135092,48131035,48130459,48131802,48131589,48133467,48130781,48148795],"score":10,"text":""Starting June 15, 2026, Agent SDK and claude -p usage on subscription plans will draw from a new monthly Agent SDK credit, separate from your interactive usage limits."

Details: https://support.claude.com/en/articles/15036540-use-the-claude-agent-sdk-with-your-claude-plan","time":1778725110,"title":"Tell HN: Starting June 15, claude -p usage will change","type":"story"}],[48148115,{"by":"domysee","descendants":0,"id":48148115,"kids":[48148458,48148120],"score":3,"time":1778850500,"title":"Update on GitHub Copilot usage-based pricing","type":"story","url":"https://github.blog/news-insights/company-news/github-copilot-individual-plans-introducing-flex-allotments-in-pro-and-pro-and-a-new-max-plan/"}],[48130526,{"by":"hsin003","descendants":1,"id":48130526,"kids":[48133309,48137435,48130577,48130530,48134990],"score":5,"time":1778726796,"title":"Show HN: Containarium – self-hosted sandbox for AI agents, MCP-native","type":"story","url":"https://github.com/footprintai/Containarium"}],[48122031,{"by":"franze","descendants":0,"id":48122031,"kids":[48122116],"score":3,"time":1778680942,"title":"Ask HN: What is better Opus 4.6 High or Opus 4.7 Medium?","type":"story"}],[48120148,{"by":"jesseduffield","descendants":0,"id":48120148,"kids":[48121422],"score":2,"text":"My colleague and I have found AI is consistently falling short on the following things. I'd be interested to know whether this is a prompting issue or a model issue (we use claude 4.7):

- comments being too verbose, and making reference to the previous code. Also just using phrases that no human would ever use (e.g. saying 'gate', 'belt-and-braces' a lot)

- over-using default arguments in method definitions

- failing to see the bigger picture in terms of DRYness, code organisation

- way too much defensive coding like handling edge cases that would never happen, adding fallbacks for everything","time":1778668568,"title":"What is your AI agent constantly doing wrong?","type":"story"}],[48071940,{"by":"pretext","descendants":274,"id":48071940,"kids":[48072400,48072225,48073789,48072075,48072471,48073497,48072293,48075079,48074926,48072214,48072197,48075652,48074807,48073090,48072681,48073811,48072515,48072970,48073830,48073049,48072061,48072224,48072077,48075335,48075635,48087370,48074483,48075629,48076752,48084297,48077528,48075567,48076609,48075578,48076505,48080955,48072527,48072467,48081932,48072542,48075221,48075332,48072826,48073019,48076636,48082973,48074109,48073213,48072184,48072254,48072107,48080948,48072701,48072811,48075367,48076922,48077694,48075813,48080889,48075684,48072352,48075683,48074309,48074761,48081488,48075260,48072637,48073540,48077073,48074160,48075739,48074708,48073943,48073501,48077500,48081658,48075706,48079298,48077559,48075139,48073289,48076209,48076041,48075983,48076128,48077345,48075803,48075690,48084208,48072368,48078487,48072280,48072123,48075849,48078190,48075883,48075570,48101217,48074706,48079500,48072661,48180259,48076921,48072649,48076560,48072175,48076430,48076733,48072933,48075909,48076004,48073992,48108604,48073528,48079441,48107774,48073686,48074308,48082043,48072429,48076893,48083132,48075992],"score":528,"text":"Examples: https://thariqs.github.io/html-effectiveness/

Related: https://simonwillison.net/2026/May/8/unreasonable-effectiven...","time":1778302432,"title":"Using Claude Code: The unreasonable effectiveness of HTML","type":"story","url":"https://twitter.com/trq212/status/2052809885763747935"}],[48089289,{"by":"cratermoon","descendants":108,"id":48089289,"kids":[48090348,48090385,48094826,48092010,48096459,48090072,48094485,48094528,48090568,48093248,48091358,48091408,48099637,48089976,48117596,48090691,48092011,48091113,48091081,48091430,48090593,48131667,48095422,48095406,48095603,48093957,48097894,48094046,48099077,48099548,48092328,48092812,48094365,48100024,48095799,48093250,48091589,48098910,48096556,48090239,48108213,48095325,48093030,48114136,48103744,48114304,48094515,48098242,48098531,48094517,48110066,48091915,48091583,48091640],"score":380,"time":1778456395,"title":"An AI coding agent, used to write code, needs to reduce your maintenance costs","type":"story","url":"https://www.jamesshore.com/v2/blog/2026/you-need-ai-that-reduces-your-maintenance-costs"}],[48023496,{"by":"mtricot","descendants":48,"id":48023496,"kids":[48025464,48029859,48025669,48029053,48036100,48032152,48028308,48033031,48034010,48032521,48023759,48023678,48027322,48031350,48130540,48036953,48045016,48027860,48031500,48073604,48030449,48028522,48123882,48057146,48080396,48066681,48027017,48024350,48034927,48079947,48024843,48064727,48029442,48028404],"score":156,"text":"$17","time":1777993398,"title":"Show HN: Airbyte Agents – context for agents across multiple data sources","type":"story"}],[48063548,{"by":"doshay","descendants":67,"id":48063548,"kids":[48065446,48064990,48067231,48065453,48067158,48067135,48065150,48066993,48064799,48064856,48065130,48064859,48066733,48125574,48070892,48075035,48066028,48065047,48071427,48067016,48095948,48064772,48065166,48065065,48067075,48066675,48091730,48065680,48065840,48066677,48065412,48065849,48077768,48065312,48130735,48075926,48066043,48132116,48182059,48069543,48092546,48072996,48070680,48063730,48064261,48064317,48065810,48067843,48072339,48073190,48068703,48085539,48064864],"score":129,"text":"hi guys.\nbeen working on something i think is fundamentally missing in today's workflow with ai agents.

vcs.

i find myself struggling with questions that agents can't answer like "why did you do it?", "when did u delete this folder? why?", etc. or trying to /rewind (after a /compact...) or basically `bisect` to find when and why something was done by the agent in the current / previous session.

just like git did for code, i think we are the same core capabilities with ai agents

so...

i developed an open source solution for that (currently supporting claude code)

would love to get feedback, contribution or maybe other ideas or solutions you find for those problems.","time":1778249736,"title":"Show HN: Git for AI Agents","type":"story","url":"https://github.com/regent-vcs/re_gent"}],[47962032,{"by":"pmig","descendants":298,"id":47962032,"kids":[48021307,48020461,48020312,48023982,48021239,48020741,48021196,48021003,48026090,48021460,48021686,48020327,48022679,48020890,48020316,48021654,48022180,48029866,48022187,48031517,48020716,48021768,48021691,48022262,48021252,48007786,48028393,48031633,48020653,48020797,48024885,48027983,48032414,48025968,48022046,48033209,48030008,48021106,48023826,48022909,48039814,48033514,48022131,48043153,48021219,48032947,48021367,48027340,48030528,48024753,48029076,48024571,48023382,48007299,48021394,48020628,48031247,48027769,48021527,48025728,48028276,48021486,48021689,48028216,48028700,48021335,48022972,48020888],"score":413,"time":1777555303,"title":"Should I run plain Docker Compose in production in 2026?","type":"story","url":"https://distr.sh/blog/running-docker-in-production/"}],[47995037,{"by":"bring-shrubbery","descendants":46,"id":47995037,"kids":[47997730,47996413,47996194,47999648,47996007,47996508,47999850,47997752,47996977,48003143,48017154,47996708,47997406,48002824,47997877,47995061,47995916,48017177,48021412,47999892,47996524,47996958,47995505,47996441,47996646],"score":185,"text":"Hi HN, author here. SHARP is Apple's recent single-image 3D Gaussian splatting model (https://arxiv.org/abs/2512.10685). Their reference code is PyTorch + a pretty heavy pipeline; I wanted to see if it could run in a browser with no server hop, so I exported the predictor to ONNX and ran it via onnxruntime-web with the WebGPU EP.

What works: drop in an image, get a .ply you can download or preview live, all on your machine — your image never leaves the tab. The model is large (~2.4 GB sidecar) so first load is slow on a cold cache, but inference itself is a few seconds on a recent Mac.

Caveats: SHARP's released weights are research-use only (Apple's model license, not the code's). I host the exported ONNX on R2 so thedemo "just works", but you can also export your own from the upstream Apple repo and upload locally.

Happy to talk about it in the comments :)","time":1777799696,"title":"Show HN: Apple's SHARP running in the browser via ONNX runtime web","type":"story","url":"https://github.com/bring-shrubbery/ml-sharp-web"}],[47990675,{"by":"shad42","descendants":121,"id":47990675,"kids":[47991503,47991481,47991871,47991683,47991108,47991000,47991419,47996350,47991486,47991796,47991669,47991972,47993677,47997993,47992009,48001088,47993487,47996437,47993579,47994866,47998078,47991208,47991347,48001671,47991477,47992502,47992777,47995564,47993353,47991214,48002289,47990998,47997618,47995867,47993958,47997498,47991142,47991128,47998321,48159555,47991058,47995623,47994979,47994031,47992814,47994647,47999058,47992077,47993129,47993142,47991147,47997631],"score":182,"time":1777756887,"title":"The agent harness belongs outside the sandbox","type":"story","url":"https://www.mendral.com/blog/agent-harness-belongs-outside-sandbox"}],[47989883,{"by":"indrora","descendants":850,"id":47989883,"kids":[47991835,47990516,47990657,47990808,47990878,47990362,47990582,47990547,47993828,47990343,47990556,47991964,47990694,47991786,47992221,47990515,47992277,47992554,47990595,47990610,47990486,47994663,47991814,47992441,47994259,47990833,47992183,47990706,47991921,47990512,47992341,47990423,47992543,47991036,47990829,47991954,47991562,47990540,47992227,47992672,47990919,47991102,47997104,47992996,47990473,47993871,48013259,47991644,47995943,47994046,47991852,47999628,47991759,47998877,47999939,47991143,47995898,47992296,47997199,47991795,47991979,47992152,47995095,47996711,47996445,47990493,47990852,47995902,47997032,47990545,47990792,47990403,47990752,47999877,48012302,47994810,47994884,47992795,47993218,47991220,47990717,47990896,47992467,47992321,47994745,47990593,47995611,47991160,47991990,47994820,47991764,47993544,47990764,47990340,47991064,47994435,47990397,47996542,47998648,47996318,47998312,47992080,47995344,47993635,47997343,47996610,47990445,47993971,47990499,47994749,47991109,47991715,47990538,47990460,47990839,47994407,47994741,48003615,47995759,48004953,47992854,47991778,48031525,47995016,47995019,47992075,47990390,48005005,47990470,47990432],"score":1513,"time":1777751846,"title":"VS Code inserting 'Co-Authored-by Copilot' into commits regardless of usage","type":"story","url":"https://github.com/microsoft/vscode/pull/310226"}],[47981669,{"by":"susam","descendants":66,"id":47981669,"kids":[47982876,47982200,47982216,47982429,47984575,47987837,47989345,47991329,47984654,47987672,47987295,47982598,47987444,47982953,47986245,47983490],"score":135,"time":1777677850,"title":"A report on burnout in open source software communities (2025) [pdf]","type":"story","url":"https://mirandaheath.website/static/oss_burnout_report_mh_25.pdf"}],[47982718,{"by":"mantiscore","descendants":3,"id":47982718,"kids":[47984363,47982917,47983513,47982719,47996321,47983393,47983488,48015645,48030904],"score":16,"time":1777688398,"title":"Governor – a Claude Code plugin to reduce token/context waste","type":"story","url":"https://github.com/0xhimanshu/governor"}]]

user:	eddyaipt
created:	April 26, 2026
karma:	6
about:	Builder focused on AI coding tools, agents, devtools, and startup workflows. Interested in practical automation, evals, and reliability. X: https://x.com/eddyaipt PH: https://www.producthunt.com/@eddyaipt

1.	1 points by eddyaipt 34 days ago \| discuss
2.	36 days ago \| discuss