Show HN: Local LLM Notepad – run a GPT-style model from a USB stick https://bit.ly/4nsvttJ

Show HN: Local LLM Notepad – run a GPT-style model from a USB stick What it is A single 45 MB Windows .exe that embeds llama.cpp and a minimal Tk UI. Copy it (plus any .gguf model) to a flash drive, double-click on any Windows PC, and you’re chatting with an LLM—no admin rights, Cloud, or network. Why I built it Existing “local LLM” GUIs assume you can pip install, pass long CLI flags, or download GBs of extras. I wanted something my less-technical colleagues could run during a client visit by literally plugging in a USB drive. How it works PyInstaller one-file build → bundles Python runtime, llama_cpp_python, and the UI into a single PE. On first launch, it memory-maps the .gguf; subsequent prompts stream at ~20 tok/s on an i7-10750H with gemma-3-1b-it-Q4_K_M.gguf (0.8 GB). Tick-driven render loop keeps the UI responsive while llama.cpp crunches. A parser bold-underlines every token that originated in the prompt; Ctrl+click pops a “source viewer” to trace facts. (Helps spot hallucinations fast.) https://bit.ly/3I89nwr July 1, 2025 at 12:43AM

Show HN: Timezone converter that tells you if your meeting time sucks https://bit.ly/3I0vLrA

Show HN: Timezone converter that tells you if your meeting time sucks I work with a team spread across Sydney, London, and SF. Last month I accidentally called my Aussie colleague at 3am their time during what I thought was a "quick sync". The silence before "mate... do you know what time it is here?" still haunts me. Built this: https://bit.ly/3I85vLN It's a timezone converter but it tells you if your meeting time sucks for the other person: - Meeting quality ratings (excellent/good/fair/poor) - Visual indicators for day/night - Shows if it's a holiday in their country - Handles weird cases like Dubai's Sunday-Thursday workweek Technical bit: pre-generated 18k+ static pages for every city combination. Loads instantly because there's no backend calculations. Next.js 15, no database. Still figuring out monetization (ads? affiliate links for virtual meeting tools?) but keeping it free for now. What else would make this useful? Currently tracking holidays for ~20 countries but could add more. https://bit.ly/3I85vLN June 30, 2025 at 10:37PM

Show HN: Cheesy Mamas: Local only code editor with Git and Bash support https://bit.ly/3Goc3Fu

Show HN: Cheesy Mamas: Local only code editor with Git and Bash support Cheesy Mamas is a local first, multi tab code editor written in Python using PyQt6. It is designed for Linux systems and built around simplicity, transparency, and control. There is no telemetry, no sync, and no accounts. The editor runs entirely on your local machine using standard system tools and stays out of your way unless you ask for help. The editor supports multiple files open at once, persistent tab state, live dirty tracking, and a dark UI. It includes syntax highlighting for Python, C, and LaTeX. A built in run button executes Python directly, compiles C with gcc, or runs pdflatex for LaTeX files. It also includes a Bash button to launch or edit a saved shell script. There is no plugin system and no background processes. All functionality is visible and inspectable in the interface. The Git integration is the core design focus. Unlike most editors, which treat Git as a sidebar or rely on an external staging panel, Cheesy Mamas embeds Git version history directly beside each open file. When you open a file, the editor checks if it is part of a Git repository. If not, the first commit you make will automatically initialize a new Git repository in the current folder. For each file, Cheesy Mamas retrieves its individual commit history using Git log limited to that path. This history appears in a vertical sidebar next to the editing pane. Selecting a commit loads that exact version of the file from Git and performs a diff against the current working version in memory. The editor highlights changed lines and overlays revert options directly into the document view. When you click a past commit, the editor compares that version against your current working file. All changed lines are visually marked. You can click a "revert line" button next to any highlighted block to immediately undo that change using the older version. These changes are local until you save. This allows for a granular, low effort recovery flow without affecting unrelated files or requiring a full diff tool. Right clicking a commit provides a context menu that lets you view the full unified diff, copy the full version of that commit to your clipboard, or revert the entire file to that point. These operations use standard Git plumbing internally and do not alter other files in the repository. Cheesy Mamas does not require you to commit or stage across all files. Each file's history and actions are isolated. The editor is single instance by default. Opening a file from the file browser or terminal reuses the existing window and opens the file in a new tab. This is handled via a relay system that passes the file path to the existing running instance. The UI is dark by default with soft gold highlights. There is no animation or decoration beyond what is needed for clarity. The editor warns on exit if any file is unsaved. Saving and Git commits are handled through dedicated buttons and keyboard shortcuts. The Bash button opens a terminal script from the config folder, or lets you write one if none exists. Cheesy Mamas was built to solve a personal problem. Most editors assume the user is syncing code to a cloud service or using Git externally. They require plugins or navigation panels to access version history and rarely show diffs in context. Cheesy Mamas was designed to treat versioning as a natural part of editing, and to bring Git history as close to the cursor as possible without overwhelming the UI. The project is fully offline, runs on Linux, and is installable via a simple shell script. It places the Python script and assets in `~/.local/share/CheesyMamas`, creates a `.desktop` entry, and integrates with your application menu. You can optionally set it as the default handler for `.py`, `.c`, `.tex`, and `.sh` files by editing the desktop file and uncommenting the `MimeType` field. There is no account system and no sync. It’s a local program, designed to live where you live, and let you undo what needs undoing. https://bit.ly/45NzkuV June 30, 2025 at 04:53AM

Show HN: BloomPilot – AI-Powered Overlay for Bloomberg Terminal https://bit.ly/4498JY7

Show HN: BloomPilot – AI-Powered Overlay for Bloomberg Terminal Hi HN, We just launched BloomPilot — a minimal AI-powered overlay designed for Bloomberg Terminal users. It's built for financial professionals who want faster GPT-enhanced insights, a lightweight terminal interface, and modern tooling on top of the Bloomberg infrastructure they already use. Key Features: GPT-4o analysis integrated into Bloomberg-style command line Built-in fallbacks (Alpha Vantage, Polygon, Finnhub) if Bloomberg API is unavailable One-time payment of 299 USDC via Phantom wallet (Solana) Terminal-style UI with keyboard-first design and command history Real-time data streaming, AI formatting, wallet-based access control It’s designed specifically for professional traders, analysts, and fintech builders who spend their day in BBG and want a smarter way to interact with it. We’re focused on performance, authenticity (BBG UI), and simplicity — no freemium models, no monthly billing, and no fluff. Would love your thoughts, questions, or ideas for features. https://bit.ly/448lGBl June 29, 2025 at 11:10PM

Show HN: AI-powered tracker of Trump executive orders https://bit.ly/4472TGD

Show HN: AI-powered tracker of Trump executive orders I built a tracker that automatically scrapes the White House website for new executive orders and uses GPT-4 to generate plain-English summaries. The system runs daily, finds new orders, feeds the full legal text to ChatGPT for summarization and auto-categorization, then generates individual pages and updates the main index. It even creates custom Open Graph images for social sharing. Currently tracking 158+ orders with automatic updates as new ones are signed. Features: - AI summaries of all executive orders in plain English - Auto-categorization by policy area (immigration, trade, AI, etc.) - Search by keyword, date, or category - Completely neutral - Individual pages for each order with full text - Auto-generated OG images I got tired of reading dense legal text to understand what's actually being signed. The AI does the heavy lifting of parsing government language into readable summaries. Link: https://bit.ly/44BCp05 Tech: Next.js/Tailwind frontend, Python scraper with BeautifulSoup, GPT-4 for summaries, automated OG image generation via headless chrome. https://bit.ly/44BCp05 June 30, 2025 at 12:51AM

Show HN: Tablr – Supabase with AI Features https://bit.ly/3GpaV4x

Show HN: Tablr – Supabase with AI Features https://bit.ly/45Oqop8 June 30, 2025 at 12:05AM

Show HN: DNS at ludicrous speed for Go, powered by XDP sockets https://bit.ly/4lQfVyr

Show HN: DNS at ludicrous speed for Go, powered by XDP sockets https://bit.ly/3GnoQYM June 29, 2025 at 07:27AM

Show HN: SVG Lined Tile Generator https://bit.ly/4nrv1vH

Show HN: SVG Lined Tile Generator https://bit.ly/4kZvQKw June 26, 2025 at 02:13AM

Show HN: Leveraging Google ADK for Cyber Threat Intelligence https://bit.ly/4kdJAA0

Show HN: Leveraging Google ADK for Cyber Threat Intelligence The project is in an early state. I just had to recently reset the graph data store, but I figured now is a good time to share my post and project. The link is to my blog post, the tool is at https://bit.ly/3Ti9hVh https://bit.ly/4lyK7Ok June 28, 2025 at 09:19PM

Show HN: AIOps MCP – Log anomaly detection using Isolation Forest https://bit.ly/3ZTGs5n

Show HN: AIOps MCP – Log anomaly detection using Isolation Forest I built an open-source AIOps MCP (Monitoring & Control Plane) that detects anomalies in logs using Isolation Forest. It accepts logs from agents, apps, or collectors, parses and extracts features, and identifies unusual patterns in real time. Alerts can be sent to Slack, Webhooks, or PagerDuty. It’s lightweight, easy to deploy with Kubernetes & Helm, and designed to plug into existing observability stacks. I built this to experiment with combining ML-based anomaly detection and flexible alerting for DevOps/SRE teams. Most AIOps platforms are either too heavyweight or closed-source — I wanted something minimal yet effective. You can try it by running the FastAPI app locally or deploying with Helm. Contributions are welcome — I’d love feedback on features, detection accuracy, and real-world use cases! GitHub: https://bit.ly/3ZSDldO https://bit.ly/3ZSDldO June 28, 2025 at 07:09AM

Show HN: Self-host your data anonymization pipeline https://bit.ly/3I4C79d

Show HN: Self-host your data anonymization pipeline Needed this in my own work, anonymizing PII/PHI and decided to build this because presidio didn't really cut it for our use-case. Try it and maybe let me know if you have any feedback :) https://bit.ly/445JwO9 June 28, 2025 at 04:01AM

Show HN: Dungeon Master in Your Console https://bit.ly/3G0AhFV

Show HN: Dungeon Master in Your Console I don't normally share side projects here(or in general). Don't have much time to open them up to too much attention. I started this project while riding in a car last weekend. Mainly to explore OpenAI Codex. Using Github mobile I wrote the initial specifications into the readme, and using the ChatGPT iOS app, had Codex build a simple CLI based dungeon master. Switched back to Github for managing the PRs and back and forth for the whole car ride... It kinda got a little out of hand from there, and it's now a mix of AI(mostly AI) and myself making adjustments... The first version was entirely OAI and it worked OK but was too easy on the player. Thanks to HN I had heard about the Wayfarer model and I find that model to be pretty entertaining. In the end I thought this turned out pretty "cute" and makes a decent time waster that looks like work wink wink https://bit.ly/3GcRzj8 June 27, 2025 at 09:58PM

Show HN: I wrote a GPU-less billion-vector DB for molecule search (live demo) https://bit.ly/4eHxXk5

Show HN: I wrote a GPU-less billion-vector DB for molecule search (live demo) Input a SMILES string (or pick one molecule from the examples) and it returns up to 100k molecules closest in 3-D shape or electrostatic similarity – from 10+ billion scale databases — typically in under 5-10 s. *Why it might interest HN* * Entire index lives on disk — no GPU at query-time, less than ~10 GB RAM total. * Built from scratch (no FAISS index / Milvus / Pinecone). * Index-build cost: one Nvidia T4 (~ 300USD) for one 5.5B database. * Open to anyone, predict ADMET, export results as CSV/SDF. Full write-up & benchmarks (DUD-E, LIT-PCBA, SVS) in the pre-print: https://bit.ly/4kbb3SW... https://bit.ly/46fhMrK June 27, 2025 at 12:51AM

Show HN: Listed – An agentic platform to rank your business on AI https://bit.ly/44AUBHc

Show HN: Listed – An agentic platform to rank your business on AI Hi HN, I’m Harrison, co-founder of Listed. Today we're launching our agentic platform to help your business win in the new age of AI. You can try the platform here: https://bit.ly/4nB9yAG And watch the launch video here: https://www.youtube.com/watch?v=MJUPo6H78z8 The idea for this came from pure frustration. I asked ChatGPT about my own company and it hallucinated, inventing features and getting basic facts wrong. I realized there was no mechanism for a business to provide a verified source of truth to these models. This problem is now existential. With Google's AI Overviews and the rise of answer engines, your website's unstructured HTML is a poor source for the rich, nuanced context that LLMs need. This leads to an army of AI bots from OpenAI, Google, Perplexity, etc., scraping your site, getting it wrong, and permanently baking those errors into their models. So, we built Listed. The simplest analogy is it's like Cursor, but for context. Instead of an AI helping you write code, our agent helps you build the comprehensive, structured context that allows LLMs to represent your business accurately and favorably. Here’s how our agentic system works: Automated Context Building: When you sign up, our agent scrapes your existing website to build a first draft of your AI Listing. It structures the data and identifies weak spots. Intelligent Workflows: Based on ongoing analytics, the agent initiates simple, chat-based workflows to help you enrich your listing and improve its accuracy and ranking potential. Performance Analytics & Feedback Loop: The agent constantly measures your AI Ranking (discoverability) and Recall Accuracy across all major models (GPT-4o, Claude 3, Gemini, etc.). This data feeds back into the system, generating new workflows to continuously improve your performance. The Connection: Your AI Listing is a hosted service. You add a simple code snippet to your website. When AI crawlers visit, this acts as a signpost, essentially "prompt injecting" and directing them to consume your clean, structured, AI-optimized data feed instead of trying to parse your messy site. The goal is to give every business an active role in the AI ecosystem. You provide the clean, verified data that AI companies desperately need, and in return, you get to control your narrative and rank higher in their answers. We are launching our free tier today. We’d love for you to try it out and hear your feedback. You can get started here: https://bit.ly/4nB9yAG I'll be here all day answering questions. Thanks! June 27, 2025 at 01:32AM

Show HN: What time is it in Corporate https://bit.ly/45DYb4n

Show HN: What time is it in Corporate Three months ago someone posted a site that showed time in corporate [1] The most interesting comment to me was about the National Retail Federation 4-5-4 calendar. That calendar was hard to understand, so I made a visualization of it. I also implemented the other calendar types people were asking for. I did find the SEC dataset of 10032 publicly traded companies [2][3], but have not finished implementing a search for symbol yet. [1] https://bit.ly/3CU0QLr [2] https://bit.ly/40e5qwv... (SUB data set, field `fye`) [3] https://bit.ly/44fQVsZ https://bit.ly/3I1QISP June 26, 2025 at 11:43PM

Show HN: AI Phone Interviewer – get a call in 30 seconds https://bit.ly/4nlPRww

Show HN: AI Phone Interviewer – get a call in 30 seconds Enter your phone number, get called in 30 seconds for a 2–3 minute AI-powered screening interview. https://bit.ly/3FVTLLR Current MVP scope Right now it handles general screening questions and generates simple reports. We’re validating demand before building: Technical screening libraries ATS integrations Custom question sets per role or company Multi-language support Who we’re looking for We’d love feedback from recruiters and startup founders who are (or soon will be) running hiring processes. Request for feedback Please actually try the call first—I know it sounds gimmicky, but the voice quality will surprise you. Then let us know: Did it feel natural? Would you be comfortable being screened this way? If you hire, could you see your team using this? What needs improvement? To see the full recruiter dashboard, leave your email on the page and we’ll send you the demo. This is just an MVP to test the concept. Curious what HN thinks—future of recruiting or unnecessary automation? June 26, 2025 at 02:50AM

Show HN: Voice-Mode MCP – Conversational Coding for Claude Code, Gemini CLI https://bit.ly/4k63mh7

Show HN: Voice-Mode MCP – Conversational Coding for Claude Code, Gemini CLI When I heard about Google's clone of Claude Code this morning I tried out my 2 week old MCP server and instantly had two way voice conversation with it. Gemini seemed a bit confused by this. :-) https://youtu.be/HC6BGxjCVnM?feature=shared&t=36 It's a FOSS MCP server I created a couple of weeks ago: - https://bit.ly/4ke7fR8 - https://bit.ly/3G83PRY # Installation (~/.gemini/settings.json) { "theme": "Dracula", "selectedAuthType": "oauth-personal", "mcpServers": { "voice-mode": { "command": "uvx", "args": [ "voice-mode" ] } } } https://bit.ly/4ke7fR8 June 26, 2025 at 01:02AM

Show HN: I built a cloud on my own ASN w real 1:1 compute to fight the cartels https://bit.ly/3TautfS

Show HN: I built a cloud on my own ASN w real 1:1 compute to fight the cartels Sup HN I'm MX, a solo founder building Infuze Cloud, launching as a beta today. I started this project because I was tired of a long list of reasons why the cartels are ridiculous that I shall not dwell on for too long here because I had to rewrite this twice cause of hitting the character limit So I decided to try building something I’d want to use. The whole stack is built custom from the ground up with no external dependencies or costs to third parties apart from hardware and IP space. What Infuze is: Raw, dedicated performance: 1 vCPU = 1 physical thread. No overcommit. I cap my nodes allocations at the physical hardware limit with some overhead. Pricing based on what you use: It's set at $10/m for 4gb/1vcpu/50gb but can be provisioned for min 1hour(3 cents). Discounts on wallet top up's to prevent pressure to get unneeded resources(even the minimum top up starts at 10% discount so its actually $8+ a month goes down to $7.50 with larger top up's. Runs on our own hardware(leased) and autonomous system: We operate our own Intel Xeon Platinum 8280 servers and BGP-routed IP space (AS211747). KVM based with storage NVMe gen4 with ZFS Stack: It's built on mostly open source technology! Proxmox for virtualization, Knot for master authorative and outsourced for anycast slaves, custom Go microservices for most of the automation needed by the frontend(open sourcing some of them soon!). FRR for BGP. Networking is standard bridged networking that's routed from leased IP space that I'm announcing with FRR. Mail using maddy. Prometheus/node exporter for metrics, grafana for panels. The LLM chatbot is using AnythingLLM with openrouter but that was mostly like a FOMO thing lol tbh I don't expect anyone to use it much(because i dont) but if it helps someone then that's great. Support/ticketing is custom, with the Next.js frontend, billing is Stripe. Each VM gets a public IPv4 and a /64 subnet routed to it, no NAT or SNAT. If you guys have any questions or want to discuss more on the stack that I didn't mention I'm very open to sharing, discussing, and learning about new ways to optimize my stack. I'm still very new to all this and learning as I go along so any insight is appreciated! I'm working on something more experimental(custom firecracker fork that directly boots ELF+IVSHMEM apps from memory with a unikernel or initramfs), which I hope to bring to the public soon, but lacked funding to keep moving forward so I decided to start this as a learning experience and first venture into the industry, with a more mature stack that's reliable and battle tested enough for public use. Who it's for: Developers who prefer using linux with root access, via SSH, etc. People who want to pay for something closer to real infra costs. Compute isn't expensive and the tech isn't difficult. We shouldn't be forced to pay an amount that a monopoly feels they deserve. This isn't for those that engage in yaml-therapy or love contributing to the charitable foundation for wooden figureheads, but I've got something lined up for you guys too! ;) This is the first public beta, and while most things are battle-tested, I expect a few bumps. I’ll be around all day to answer any questions, fix bugs quickly, and learn from the feedback. For the benchmark nerds I spun up a quick little site for fun with v0 because I was finding endless things to tinker with and feed my impostor syndrome to delay launching this but I've dragged it long enough https://bit.ly/442jz23 You can get a free dollar voucher there and run a benchmark for fun, I wanted to do a larger amount but realized that my hourly billing is gonna be a magnet for abuse and being entirely self funded that's probably not a good idea, but I'm prepared to pivot fast(and strike back, to those even considering it -_-) Thanks for reading, and I’d love to hear any feedback, ideas, and critique. Appreciate you all. https://bit.ly/4lsyecI June 26, 2025 at 12:58AM

Show HN: Linux tool to save and recall memories quickly https://bit.ly/40k3PVI

Show HN: Linux tool to save and recall memories quickly i made this small Linux command tool which works by managing simple quick things you want to remember like a password/command/ideas, things that you might just need them noted somewhere. it works by allowing you to make files that work like categories each file has his own list of memories, each memory has it's own keywords, which are used to search for that memory when you need it. this project was made for 2 reasons, first was a personal need for a tool that remembers certain linux commands i use in like once a while, i know i could have just searched around for some tools that work this way, but i wanted to take this project as a way to learn shell scripting with no third party libraries, so it should be work natively on most linux distros. i would like contributions and ever more if they contain comments. https://bit.ly/4loEwKc June 25, 2025 at 11:34PM

Show HN: iCloud Drive Sync on Windows will not progress and how I fixed it https://bit.ly/469Z9p4

Show HN: iCloud Drive Sync on Windows will not progress and how I fixed it I noticed my iCloud drive was not synchronizing. I fixed it then it happened again a week or two later. I decided to write a blog post of the steps to fix it for myself or others as all steps I found online did not work and I wasted a lot of time. https://bit.ly/4niOqPe June 25, 2025 at 07:22AM