15:[[47678844,{"by":"manzt","descendants":35,"id":47678844,"kids":[47718992,47723579,47725875,47717985,47680314,47719711,47718741,47717386,47721781,47719257,47776368,47720403,47722750,47719658,47719823,47718300,47722364,47679117,47729013],"score":140,"text":"$16","time":1775584066,"title":"Show HN: Marimo pair – Reactive Python notebooks as environments for agents","type":"story","url":"https://github.com/marimo-team/marimo-pair"}],[47554904,{"by":"supermdguy","descendants":0,"id":47554904,"kids":[47555018],"score":11,"time":1774707906,"title":"You can't imitation-learn how to continual-learn","type":"story","url":"https://www.lesswrong.com/posts/9rCTjbJpZB4KzqhiQ/you-can-t-imitation-learn-how-to-continual-learn"}],[47554319,{"by":"nomoreaccts","descendants":12,"id":47554319,"kids":[47554867,47554644,47554980,47554462,47554382,47554968,47556935,47554989,47554781,47555007,47554752,47554769,47556538,47555460,47554380],"score":4,"text":"All new user comments are now hidden from public view.

What’s this gonna do to the platform?","time":1774703850,"title":"Ask HN: HN now hides comments from new users","type":"story"}],[47515381,{"by":"akadeb","descendants":4,"id":47515381,"kids":[47521519,47521395,47516185,47521676,47517137],"score":8,"text":"CharacterAI has over 20M active users and lets people talk to AI voice clones. I built a local alternative that clones any voice with <10s of audio and can talk fluently in over 10 languages, and runs locally on your Macbook or Mac mini.

The voice pipeline currently supports MLX on any M1 through M5 chip. I used Whisper-Turbo for STT, Qwen3.5-9B-4bit for the LLM and Qwen3-TTS-0.6B-4bit for TTS.

The repo also has a Websocket Transport to add these voices to devices powered by the ESP32 via secure websockets.","time":1774433243,"title":"Show HN: I built a free CharacterAI that runs locally","type":"story","url":"https://github.com/akdeb/open-toys/tree/main/resources"}],[47515351,{"by":"ElFitz","descendants":2,"id":47515351,"kids":[47515595,47515385],"score":2,"text":"$17","time":1774433027,"title":"Show HN: Lightless Labs Refinery – multi-model consensus and synthesis","type":"story","url":"https://github.com/Lightless-Labs/refinery"}],[47511823,{"by":"makingstuffs","descendants":12,"id":47511823,"kids":[47515304,47520714,47518294,47521492,47516515,47513599,47512151,47512401,47512404,47512189,47520326,47547176,47515005,47515197,47529807,47518575,47528169,47512261,47515368,47513029,47515329,47513727,47512073,47512357],"score":13,"text":"$18","time":1774400427,"title":"Ask HN: How do you offload all coding to AI?","type":"story"}],[47490244,{"by":"sayYayToLife","descendants":8,"id":47490244,"kids":[47490563,47493066,47493561,47491233,47500613,47502450,47515214,47496042,47518422,47501806,47512688,47493780,47490456],"score":7,"text":"As I use open claw I am concerned about prompt injection more than bad code. However I think I'm irrationally paranoid. I'm small fries I'm a single individual, someone actively trying to hack me is exploiting a multi-million dollar zero day and they're not doing that on me.

If we're thinking about accidentally deleting everything through a terminal command, I've yet to see this actually occur.

Deductively I can see all of the worst case scenarios with open claw. Inductively I've never seen it actually happen.

I find it a bit irrational to pretend that open claw is a genuine security risk.

The moment I see on Hacker News that someone got prompt injected, I think I'll be concerned. Until then I would need almost a lottery like chance to get hacked as the first person through prompt injection.","time":1774277011,"title":"Ask HN: If there has been no prompt injection, is it safe?","type":"story"}],[47486287,{"by":"obilgic","descendants":19,"id":47486287,"kids":[47487427,47487176,47486798,47488017,47487773,47489463,47487463,47491539,47495913,47487491,47488337,47507594,47491214,47490260,47486967,47491042,47488832,47486453,47487843,47487579,47487903,47487929,47490289],"score":44,"time":1774249602,"title":"Show HN: Agent Kernel – Three Markdown files that make any AI agent stateful","type":"story","url":"https://github.com/oguzbilgic/agent-kernel"}],[47469175,{"by":"lucaas-d3v","descendants":3,"id":47469175,"kids":[47469180,47495851,47470028],"score":3,"time":1774114296,"title":"Show HN: Flint – A compiled language to replace complex Bash scripts","type":"story","url":"https://useflint.vercel.app/"}],[47467671,{"by":"kilgarenone","descendants":42,"id":47467671,"kids":[47470447,47471053,47473009,47480621,47474090,47471955,47471235,47470652,47476352,47470290,47470115,47470067,47472787,47485025,47472436,47475821,47470951,47470035,47470068,47473026,47470938,47469980,47470638],"score":54,"text":"$19","time":1774105350,"title":"Show HN: Joonote – A note-taking app on your lock screen and notification panel","type":"story","url":"https://joonote.com/"}],[47396566,{"by":"mjcbeckett","descendants":5,"id":47396566,"kids":[47396586,47397314,47399430,47397645,47399579],"score":7,"text":"$1a","time":1773651641,"title":"Show HN: Signbee – An API that lets AI agents send documents for signature","type":"story","url":"https://signb.ee"}],[47396663,{"by":"Nesco","descendants":4,"id":47396663,"kids":[47399350,47400454,47397542],"score":5,"text":"Hey everyone, I was wondering if a Poincaré disk would be able to usefully show the local structure of a graph while showing most of one

I used a collection of ML papers to explore this","time":1773652526,"title":"Show HN: Paperverse, an Hyperbolic Presentation","type":"story","url":"https://paperverse.net"}],[47390228,{"by":"jmilinovich","descendants":8,"id":47390228,"kids":[47394806,47390311,47412999,47393986,47416999,47396346,47397094],"score":31,"time":1773598946,"title":"Show HN: Goal.md, a goal-specification file for autonomous coding agents","type":"story","url":"https://github.com/jmilinovich/goal-md"}],[47394797,{"by":"MikeLuLu","descendants":2,"id":47394797,"kids":[47402419,47396331],"score":10,"text":"Hi HN,

I’m Ethan, an indie developer.

I listen to a lot of podcasts while coding or commuting, and I often want to save interesting insights from episodes. I tried tools like MacWhisper for transcription, but it only works on macOS and the workflow didn’t quite fit what I wanted.

So I built CastLoom Pro.

It’s a desktop app that lets you search, play, download, transcribe, translate, and archive podcasts in one place. The idea is to turn podcasts into something searchable and reusable instead of just passive listening.

Some details:

- Search podcasts from Apple Podcasts\n- Batch download episodes\n- Local transcription using Faster-Whisper\n- Optional translation with DeepL or OpenAI\n- Export to Markdown / SRT / TXT / JSON\n- Send transcripts to Notion or Obsidian

Everything runs locally, so audio and transcripts stay on your machine.

It currently supports Windows and macOS.

I’d love to hear your feedback or suggestions.","time":1773630711,"title":"Show HN: CastLoom Pro – Turn podcasts into a personal knowledge base","type":"story","url":"https://castloompro.toolooz.com/?lang=en"}],[47388974,{"by":"ayoubdrissi","descendants":0,"id":47388974,"kids":[47389466],"score":4,"text":"$1b","time":1773592158,"title":"Show HN: Lengpal – simple video chat for language exchange","type":"story","url":"https://www.producthunt.com/products/lengpal"}],[47386581,{"by":"mapldx","descendants":31,"id":47386581,"kids":[47389937,47419076,47389075,47387266,47387922,47390638,47388488,47388264,47390026,47400757,47389110,47388155,47389185,47388220,47388501,47394983,47386792,47398280,47416909,47388322,47387877],"score":123,"text":"$1c","time":1773576060,"title":"Show HN: Signet – Autonomous wildfire tracking from satellite and weather data","type":"story","url":"https://signet.watch"}],[47377736,{"by":"maxilevi","descendants":0,"id":47377736,"kids":[47378237],"score":11,"text":"With the current inflow of LLM aided software, I thought I would share a cool "hand-coded" project from the previous era (I wrote this in highschool so roughly ~8 years ago).

Hedra is an open world 3d game written from scratch using only OpenGL and C#. It has quite a few cool features like infinite generation, skinned animated mesh rendering, post processing effects, etc. Originally the physics engine was also written from scratch but i swapped for the more reliable bulletphysics.","time":1773502546,"title":"Show HN: Hedra – an open-world 3D game I wrote from scratch before LLMs","type":"story","url":"https://github.com/maxilevi/project-hedra"}],[47377896,{"by":"walky","descendants":1,"id":47377896,"kids":[47378327,47377988],"score":2,"time":1773503557,"title":"Show HN: TheDayAfter – open-source addiction recovery tracker","type":"story","url":"https://thedayafter.app/?o=hn"}],[47375880,{"by":"octopusos","descendants":2,"id":47375880,"kids":[47375896,47375881,47376070],"score":3,"time":1773490441,"title":"OctopusOS – an operating system for governed AI agents","type":"story","url":"https://octopusos.ai/"}],[47365851,{"by":"ndxone","descendants":1,"id":47365851,"kids":[47365879],"score":2,"text":"https://github.com/ndx-technologies/go-apple-ads

this is super helpful to us, I think more people should be aware of this approach

combine this with your: AI agent, cronjob, and git repo of your configs — and you get automated highly efficient, fully traceable, decision log, with ability to roll-back, understand every decision, and get learning loop of AI to optimize your campaigns.

this tools work well with humans in the loop too (color formatting), and built with AI agents running in CLI in mind (no-color, filters, help pages and self-discovery). inspired by Go linters (analysers) there are common linters to detect mistakes in configuration as well.

enjoy!","time":1773416040,"title":"Show HN: Apple Ads Toolkit – Run Your Apple Ads as GitOps Terraform","type":"story"}],[47365793,{"by":"dheerusha_","descendants":3,"id":47365793,"kids":[47366018,47365852],"score":4,"time":1773415799,"title":"Show HN: Agent wrapped, like Spotify wrapped","type":"story","url":"https://swarmpack.github.io/landing-page/"}],[47363400,{"by":"jsontwikkeling","descendants":22,"id":47363400,"kids":[47365377,47364981,47366224,47364445,47364378,47364140,47364952,47364754],"score":65,"text":"$1d","time":1773404111,"title":"Show HN: Algorithms and Data Structures in TypeScript – Free Book (~400 Pages)","type":"story","url":"http://amoilanen.github.io/Algorithms-with-Typescript/"}],[47354008,{"by":"photobombastic","descendants":9,"id":47354008,"kids":[47359552,47354149,47357243,47363319],"score":12,"text":"$1e","time":1773335301,"title":"Show HN: PipeStep – Step-through debugger for GitHub Actions workflows","type":"story","url":"https://github.com/Photobombastic/pipestep"}],[47350240,{"by":"JitseLambrichts","descendants":2,"id":47350240,"kids":[47462352,47350364,47353071,47351561],"score":8,"text":"I built a local-first UI that adds two reasoning architectures on top of small models like Qwen, Llama and Mistral: a sequential Thinking Pipeline (Plan → Execute → Critique) and a parallel Agent Council where multiple expert models debate in parallel and a Judge synthesizes the best answer. No API keys, zero .env setup — just pip install multimind. Benchmark on GSM8K shows measurable accuracy gains vs. single-model inference.","time":1773321988,"title":"Show HN: Run an Agent Council of LLMs that debate and synthesize answers","type":"story","url":"https://github.com/JitseLambrichts/MultiMind-AI"}]]

user:	BloodAndCode
created:	March 10, 2026
karma:	2