🛠️ Hacker News Tools

Quadrupling code performance with a "useless" if

Show HN: Loot Raiders – an ARC Raiders-inspired inventory game in Svelte

Show HN: Self-hosted voice AI agent for Asterisk/FreePBX Hi HN folks ! I am the author of AVA, a self hosted AI Voice Agent that plugs into Asterisk&#x2F;Freepbx so you own all the aspects of an AI Voice agent in your own infrastructure. It uses Asterisk native Audiosocket&#x2F;RTP with python engine to run STT,LLM and TTS loop. The project support several full providers openai, gemini, grok, elevenlabs out of the box and also provides options to build custom pipelines by choosing different stt tts and llm.<p>It also supports full local agent if you have a GPU with 25GB RAM which enables realtime conversation along with tool calling.<p>I started this as a hobby project last year when I started exploring voice agents and every saw every Saas tried to lock you in their eco system. Since then project has taken off and a lot of asterisk people started using it due to ease of setup and wide range of providers and flexibility.<p>Core is MIT: <a href="https:&#x2F;&#x2F;github.com&#x2F;hkjarral&#x2F;AVA-AI-Voice-Agent-for-Asterisk" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;hkjarral&#x2F;AVA-AI-Voice-Agent-for-Asterisk</a> Free and open source forever.<p>No install Mock UI us hosted at: <a href="https:&#x2F;&#x2F;demo.agent6789.com" rel="nofollow">https:&#x2F;&#x2F;demo.agent6789.com</a><p>I am also exploring to convert this into a more polished version for Operators to handle multiple customer and multi tenant environment and I would genuinely like to hear feedback on it.<p>Happy to answer any questions and gather any feedback. Thanks.

Circular Obstacle Pathfinding (2017)

Show HN: Codebase Posters – turn any Git repo into generative poster art npx codebase-posters inside any git repo opens a local exhibition: 18 pieces painted live from your commit history.<p>everything renders from git log, nothing leaves your machine.

Show HN: Capn-hook for coding agents – don't grep the same mystery twice

Show HN: Agent-run – Run a coding agent in a sandboxed environment

Irish datacenters now guzzle 23% of the country's electricity

Neocities: Create your own free website

Show HN: Juggler – an open-source GUI coding agent, by the creator of JUCE Hello HN, I don&#x27;t post on here much, but wanted to get some eyes on a new project I&#x27;m just launching. I think we definitely need one more AI code agent..<p>I&#x27;m a long-term C++ dev, and over 30+ years I&#x27;ve created some successful audio dev tools (JUCE, the Tracktion DAW, the Cmajor DSP language). All of these came from me getting annoyed with something I had to use, and deciding to have a go at my own take on whatever it was.<p>So Juggler is my attempt at an AI code agent, after spending too many hours loving what the models could do, but hating the CLI experience, and having some opinions of what a better UX might be for this stuff.<p>Lots more blurb on the website and github, but a quick tech dump which might grab your attention if you&#x27;re into these things:<p>A session is a document, not a log file. Each conversation is a Yjs CRDT tree. It can branch into sub-threads (recursively), and you can drill down, backtrack, edit, undo&#x2F;redo, and inspect everything: tool calls, approvals, and the raw context JSON going to the model, etc. The UI is based around Finder-style Miller columns rather than a big doom-scroll, and is quick to navigate.<p>Because it&#x27;s a CRDT behind a local web server, multiple clients can attach P2P to a live session: the native desktop app, a browser tab, or your phone. Run the headless server on the box where the code lives, view it from wherever.<p>Almost everything is a JavaScript plugin: every item in the context (read&#x2F;write&#x2F;bash&#x2F;etc.), the LLM loop strategies, slash commands, and their UIs. You can inspect, fork, or replace any of them. I don&#x27;t do much agent customisation myself, but lots of people do, and I&#x27;d love to see what they think of with this plugin API.<p>Go backend, Wails for windowing (no Electron), plain type-checked JS (strict JSDoc), Yjs for the documents. Usual BYOK provider support: Claude (CLI or API), OpenAI&#x2F;Codex, Gemini, Ollama, OpenRouter, DeepSeek, etc.<p>The app&#x27;s AGPLv3; the extension SDK and bundled extensions are Apache-2.0, so extensions have no copyleft strings attached. No signup, no telemetry, trying to make it frictionless for people to try it out..<p>It&#x27;s very much a beta, and is a one-man side project. It hasn&#x27;t yet had a proper kicking from the real world, but I&#x27;m confident some people with similar preferences to my own will like it!<p><a href="https:&#x2F;&#x2F;juggler.studio" rel="nofollow">https:&#x2F;&#x2F;juggler.studio</a>

Claude Code sends 33k tokens before reading the prompt; OpenCode sends 7k This started based off of a hunch. We usually use OpenCode, but were &#x27;forced&#x27; to use Claude Code for a while due to issues with Meridian. In that time, we saw the usage meter rise much, much more quickly than when using OpenCode.<p>This was the initial anecdotal evidence, but we undertook this small study to collect empirical data:<p>We added logging between the agentic coding tool (Claude Code and OpenCode) and Anthropic&#x27;s endpoint, and captured all requests (and the returned usage blocks).<p>With one caveat (toward the end of the post) we found unambiguously that Claude Code was far more inefficient in terms of its cache strategy and its harness token usage than OpenCode.

An orbiting disco ball gave Einstein's theory its most precise test yet

LARP – Revenue infrastructure for serious founders

Show HN: Shirei, cross-platform GUI framework in native Go

Automation Without Understanding

The shingles vaccine may reduce the risk of dementia <a href="https:&#x2F;&#x2F;archive.ph&#x2F;PzPop" rel="nofollow">https:&#x2F;&#x2F;archive.ph&#x2F;PzPop</a>

Show HN: Introducing: Zotfile Agents Package agent behavior, skills, requirements, and permissions into a portable .zot file.

Show HN: Kurvengefahr – browser CAD/CAM for pen plotters A few years ago I made a pen plotter attachment for Prusa MK4 (<a href="https:&#x2F;&#x2F;www.printables.com&#x2F;model&#x2F;827264-pen-plotter-attachment-for-prusa-mk4" rel="nofollow">https:&#x2F;&#x2F;www.printables.com&#x2F;model&#x2F;827264-pen-plotter-attachme...</a>) and at the time I didn&#x27;t have a good way to turn artwork into G-code for it, and I put the project on ice for a while.<p>I recently wanted to dabble in line art again and made a small browser app to make it easier. As agentic AI tools of 2026 are quite addictive, it rather quickly grew into something quite a bit more - an integrated browser CAD&#x2F;CAM for pen plotters that covers everything from importing existing artwork, creating artwork from scratch, preparing for plotting and hardware integration. It includes some off-beat features like a Logo interpreter for turtle art and Graves RNN for handwriting synthesis and in addition to 3D printer pretending to be pen plotters it now also supports actual pen plotters based on EBB (AxiDraw) and GRBL firmwares through Web Serial.<p>If you own an AxiDraw or a GRBL plotter, I&#x27;d very much appreciate it you gave it a try and give feedback. As I don&#x27;t own those, I did all the testing with a hardware mock on STM32, so I am not sure how well it works attached to an actual plotter.<p>Source code and docs are on GitHub: <a href="https:&#x2F;&#x2F;github.com&#x2F;tibordp&#x2F;kurvengefahr" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;tibordp&#x2F;kurvengefahr</a>

Show HN: Skillscript – A declarative, sandboxed language for tool orchestration Hi HN — I&#x27;m Scott. Skillscript is a small language I built to write what I want my local agent to actually do, in a form I can read and version, instead of hoping the model gets it right each time.<p>The itch started with something small. I wanted my NanoClaw agent to run my morning brief the same way every day. Check overnight tickets, summarize the deploy pipeline, flag anything urgent. Every session, it would re-figure out how to do this from scratch, drift a little, and cost tokens for what&#x27;s basically a fixed procedure. I could put it in a system prompt or an MD skill file, but those are still instructions the model reads and reasons about every time. And I wanted it to run autonomously and then hand it to the model to reason over the data.<p>The second thing that pushed me: I wanted to use small local models for the cheap stuff. They&#x27;re capable, but if you just hand them the wheel, they wander. What I wanted was a way for the frontier model (or me) to write a specific procedure and hand it to the local model to execute, not interpret. The skillscript is the program; the model is the runtime.<p>Skillscript is that. A skillscript is a text file with named steps, variables, conditions, and calls out to tools (MCP connectors, a local model, and shell commands from an operator allowlist). It&#x27;s deliberately minimal — no eval, no arbitrary imports, no subprocess, no unbounded loops. Bounded language, limited potential for damage. Everything a skillscript can do is in the file. You read it and know.<p>Where it is: pre-1.0 (0.30), MCP-native, self-hosted. Rough edges I know about: first-run setup takes more steps than it should, some of the grammar is still moving, and the local model integration currently assumes Ollama. It works well enough that I use it every day, but I wouldn&#x27;t necessarily call it production-ready.<p>- Repo: [<a href="https:&#x2F;&#x2F;github.com&#x2F;sshwarts&#x2F;skillscript" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;sshwarts&#x2F;skillscript</a>](<a href="https:&#x2F;&#x2F;github.com&#x2F;sshwarts&#x2F;skillscript" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;sshwarts&#x2F;skillscript</a>)<p>- Site: [<a href="https:&#x2F;&#x2F;skillscript.ai" rel="nofollow">https:&#x2F;&#x2F;skillscript.ai</a>](<a href="https:&#x2F;&#x2F;skillscript.ai" rel="nofollow">https:&#x2F;&#x2F;skillscript.ai</a>)<p>- Docs: [<a href="https:&#x2F;&#x2F;skillscript.mintlify.app&#x2F;docs" rel="nofollow">https:&#x2F;&#x2F;skillscript.mintlify.app&#x2F;docs</a>](<a href="https:&#x2F;&#x2F;skillscript.mintlify.app&#x2F;docs" rel="nofollow">https:&#x2F;&#x2F;skillscript.mintlify.app&#x2F;docs</a>)<p>- npm: `skillscript-runtime`<p>I&#x27;d welcome critique on two things especially: the language design (is it too small? too big? wrong shape?) and the trust model around agent-authored skills. What would you want to see before you trusted this on your own machine?

[CLI Tool] Command Line Interface Guidelines