Show HN: I used Claude Code to discover connections between 100 books https://ift.tt/6jOoiuZ

Show HN: I used Claude Code to discover connections between 100 books I think LLMs are overused to summarise and underused to help us read deeper. I built a system for Claude Code to browse 100 non-fiction books and find interesting connections between them. I started out with a pipeline in stages, chaining together LLM calls to build up a context of the library. I was mainly getting back the insight that I was baking into the prompts, and the results weren't particularly surprising. On a whim, I gave CC access to my debug CLI tools and found that it wiped the floor with that approach. It gave actually interesting results and required very little orchestration in comparison. One of my favourite trail of excerpts goes from Jobs’ reality distortion field to Theranos’ fake demos, to Thiel on startup cults, to Hoffer on mass movement charlatans ( https://ift.tt/4e8AHdo ). A fun tendency is that Claude kept getting distracted by topics of secrecy, conspiracy, and hidden systems - as if the task itself summoned a Foucault’s Pendulum mindset. Details: * The books are picked from HN’s favourites (which I collected before: https://ift.tt/G4IM3Lj ). * Chunks are indexed by topic using Gemini Flash Lite. The whole library cost about £10. * Topics are organised into a tree structure using recursive Leiden partitioning and LLM labels. This gives a high-level sense of the themes. * There are several ways to browse. The most useful are embedding similarity, topic tree siblings, and topics cooccurring within a chunk window. * Everything is stored in SQLite and manipulated using a set of CLI tools. I wrote more about the process here: https://ift.tt/HqiBnbu I’m curious if this way of reading resonates for anyone else - LLM-mediated or not. https://ift.tt/7RfTOSW January 10, 2026 at 10:26PM

Show HN: Various shape regularization algorithms https://ift.tt/YSVNgyd

Show HN: Various shape regularization algorithms Shape regularization is a technique used in computational geometry to clean up noisy or imprecise geometric data by aligning segments to common orientations and adjusting their positions to create cleaner, more regular shapes. I needed a Python implementation so started with the examples implemented in CGAL then added a couple more for snap and joint regularization and metric regularization. https://ift.tt/dKZsLDi January 9, 2026 at 07:43AM

Show HN: Turn your PRs into marketing updates https://ift.tt/awf7Ev3

Show HN: Turn your PRs into marketing updates https://personabox.app January 9, 2026 at 01:05AM

Show HN: Pydantic-AI-stream – Structured event streaming for pydantic-AI agents https://ift.tt/EkqpNQK

Show HN: Pydantic-AI-stream – Structured event streaming for pydantic-AI agents https://ift.tt/wPDXhbS January 8, 2026 at 11:31PM

Show HN: I visualized the entire history of Citi Bike in the browser https://ift.tt/6VUyFx3

Show HN: I visualized the entire history of Citi Bike in the browser Each moving arrow represents one real bike ride out of 291 million, and if you've ever taken a Citi Bike before, you are included in this massive visualization! You can search for your ride using Cmd + K and your Citi Bike receipt, which should give you the time of your ride and start/end station. Everything is open source: https://ift.tt/bI7nYE4 Some technical details: - No backend! Processed data is stored in parquet files on a Cloudflare CDN, and queried directly by DuckDB WASM - deck.gl w/ Mapbox for GPU-accelerated rendering of thousands of concurrent animated bikes - Web Workers decode polyline routes and do as much precomputation as possible off the main thread - Since only (start, end) station pairs are provided, routes are generated by querying OSRM for the shortest path between all 2,400+ station pairs https://bikemap.nyc/ January 8, 2026 at 12:27AM

Show HN: bikemap.nyc – visualization of the entire history of Citi Bike https://ift.tt/kijJAMV

Show HN: bikemap.nyc – visualization of the entire history of Citi Bike Each moving arrow represents a real bike ride. There are 291 million rides in total, covering 12 years of history from June 2013 to December 2025, based on public data published by Lyft. If you've ever taken a Citi Bike ride before, you are included in this massive visualization! You can search for your ride using Cmd + K and your Citi Bike receipt, which should give you the time of your ride and start/end station. Some technical details: - No backend! Processed data is stored in parquet files on a CDN, and queried directly by DuckDB WASM - deck.gl w/ Mapbox for GPU-accelerated rendering of thousands of concurrent animated bikes - Web Workers decode polyline routes and do as much precomputation as possible off the main thread - Since only (start, end) station pairs are provided, routes are generated by querying OSRM for the shortest path between all 2,400+ station pairs Legend: - Blue = E-Bike - Purple = Classic Bike - Red = Bike docked - Green = Bike unlocked https://ift.tt/TEnKoX6 January 8, 2026 at 02:15AM

Show HN: Seapie – a Python debugger where breakpoints drop into a REPL https://ift.tt/xzg9pRN

Show HN: Seapie – a Python debugger where breakpoints drop into a REPL https://ift.tt/sh6OM0j January 7, 2026 at 11:28PM

Show HN: llmgame.ai – The Wikipedia Game but with LLMs https://ift.tt/lc34gHv

Show HN: llmgame.ai – The Wikipedia Game but with LLMs I used to play the Wikipedia Game in high school and had an idea for applying the same mechanic of clicking from concept to concept to LLMs. Will post another version that runs with an LLM entirely in the browser soon, but for now, please enjoy as long as my credits last... Warning: the LLM does not always cooperate https://www.llmgame.ai January 6, 2026 at 08:37AM

Show HN: CLI for internet speed test via Cloudflare with optional TUI https://ift.tt/4QZjOf9

Show HN: CLI for internet speed test via Cloudflare with optional TUI I've been having a lot of issues with my upload speed and wanted a cli tool to run in a cron job and couldn't quite find one to suit my needs. Sharing here in case others find it useful. https://ift.tt/mVvZF5B January 6, 2026 at 11:46PM

Show HN: WOLS – Open standard for mushroom cultivation tracking https://ift.tt/jcaCZ6S

Show HN: WOLS – Open standard for mushroom cultivation tracking I built an open labeling standard for tracking mushroom specimens through their lifecycle (from spore/culture to harvest). v1.1 adds clonal generation tracking (distinct from filial/strain generations) and conforms to JSON-LD for interoperability with agricultural/scientific data systems. Spec (CC 4.0): https://ift.tt/nuRJgHW Client libraries (Apache 2.0): Python + CLI: pip install wols (also on GHCR) TypeScript/JS: npm install @wemush/wols Background: Mycology has fragmented data practices (misidentified species, inconsistent cultivation logs, no shared vocabulary for tracking genetics across generations). This is an attempt to fix that. Looking for feedback from anyone working with biological specimen tracking, agricultural data systems, or mycology. https://ift.tt/NaRkZ3A January 6, 2026 at 12:00AM

Show HN: Unicode cursive font generator that checks cross-platform compatibility https://ift.tt/JAcbGU8

Show HN: Unicode cursive font generator that checks cross-platform compatibility Hi HN, Unicode “cursive” and script-style fonts are widely used on social platforms, but many of them silently break depending on where they’re pasted — some render as tofu, some get filtered, and others display inconsistently across platforms. I built a small web tool that explores this problem from a compatibility-first angle: Instead of just converting text into cursive Unicode characters, the tool: • Generates multiple cursive / script variants based on Unicode blocks • Evaluates how safe each variant is across major platforms (Instagram, TikTok, Discord, etc.) • Explains why certain Unicode characters are flagged or unstable on specific platforms • Helps users avoid styles that look fine in one app but break in another Under the hood, it’s essentially mapping Unicode script characters and classifying them based on known platform filtering and rendering behaviors, rather than assuming “Unicode = universal.” This started as a side project after repeatedly seeing “fancy text” fail unpredictably in real usage. Feedback, edge cases, or Unicode quirks I may have missed are very welcome. https://ift.tt/Ck1Dezj January 1, 2026 at 07:37PM

Show HN: I made R/place for LLMs https://ift.tt/02nfDrS

Show HN: I made R/place for LLMs I built AI Place, a vLLM-controlled pixel canvas inspired by r/place. Instead of users placing pixels, an LLM paints the grid continuously and you can watch it evolve live. The theme rotates daily. Currently, the canvas is scored using CLIP ViT-B/32 against a prompt (e.g., Pixelart of ${theme}). The highest-scoring snapshot is saved to the archive at the end of each day. The agents work in a simple loop: Input: Theme + image of current canvas Output: Python code to update specific pixel coordinates + One word description Tech: Next.js, SSE realtime updates, NVIDIA NIM (Mistral Large 3/GPT-OSS/Llama 4 Maverick) for the painting decisions Would love feedback! (or ideas for prompts/behaviors to try) https://art.heimdal.dev January 5, 2026 at 01:20AM

Show HN: Hover – IDE style hover documentation on any webpage https://ift.tt/3EiPRtx

Show HN: Hover – IDE style hover documentation on any webpage I thought it would be interesting to have ID style hover docs outside the IDE. Hover is a Chrome extension that gives you IDE style hover tooltips on any webpage: documentation sites, ChatGPT, Claude, etc. How it works: - When a code block comes into view, the extension detects tokens and sends the code to an LLM (via OpenRouter or custom endpoint) - The LLM generates documentation for tokens worth documenting, which gets cached - On hover, the cached documentation is displayed instantly A few things I wanted to get right: - Website permissions are granular and use Chrome's permission system, so the extension only runs where you allow it - Custom endpoints let you skip OpenRouter entirely – if you're at a company with its own infra, you can point it at AWS Bedrock, Google AI Studio, or whatever you have Built with TypeScript, Vite, and the Chrome extension APIs. Coming to the Chrome Web Store soon. Would love feedback on the onboarding experience and general UX – there were a lot of design decisions I wasn't sure about. Happy to answer questions about the implementation. https://ift.tt/pGfSeEa January 5, 2026 at 12:13AM

Show HN: ZELF – A modular ELF64 packer with 22 vintage and modern codecs https://ift.tt/w387ZVj

Show HN: ZELF – A modular ELF64 packer with 22 vintage and modern codecs https://ift.tt/wx4YtM9 January 4, 2026 at 12:59AM

Show HN: A New Year gift for Python devs–My self-healing project's DNA analyzer https://ift.tt/fwN7HDn

Show HN: A New Year gift for Python devs–My self-healing project's DNA analyzer I built a system that maps its own "DNA" using AST to enable self-healing capabilities. Instead of a standard release, I’ve hidden the core mapping engine inside a New Year gift file in the repo for those who like to explore code directly. It’s not just a script; it’s the architectural vision behind Ultra Meta. Check the HAPPY_NEW_YEAR.md file for the source https://ift.tt/jlf9tVO January 4, 2026 at 12:50AM

Show HN: Go-Highway – Portable SIMD for Go https://ift.tt/wg2KbVl

Show HN: Go-Highway – Portable SIMD for Go Go 1.26 adds native SIMD via GOEXPERIMENT=simd. This library provides a portability layer so the same code runs on AVX2, AVX-512, or falls back to scalar. Inspired by Google's Highway C++ library. Includes vectorized math (exp, log, sin, tanh, sigmoid, erf) since those come up a lot in ML/scientific code and the stdlib doesn't have SIMD versions. algo.SigmoidTransform(input, output) Requires go1.26rc1. Feedback welcome. https://ift.tt/kuZRFWK January 3, 2026 at 04:06AM

Show HN: Fluxer – open-source Discord-like chat https://ift.tt/ikEtBv8

Show HN: Fluxer – open-source Discord-like chat Hey HN, and happy new year! I'm Hampus Kraft [1], a 22-year-old software developer nearing completion of my BSc in Computer Engineering at KTH Royal Institute of Technology in Sweden. I've been working on Fluxer on and off for about 5 years, but recently decided to work on it full-time and see how far it could take me. Fluxer is an open source [2] communication platform for friends, groups, and communities (text, voice, and video). It aims for "modern chat app" feature coverage with a familiar UX, while being developed in the open and staying FOSS (AGPLv3). The codebase is largely written in TypeScript and Erlang. Try it now (no email or password required): https://ift.tt/bUf5CXl – this creates an "unclaimed account" (date of birth only) so you can explore the platform. Unclaimed accounts can create/join communities but have some limitations. You can claim your account with email + password later if you want. I've developed this solo , with limited capital from some early supporters and testers. Please keep this in mind if you find what I offer today lacking; I know it is! I'm sharing this now to find contributors and early supporters who want to help shape this into the chat app you actually want. ~~~ Fluxer is not currently end-to-end encrypted, nor is it decentralised or federated. I'm open to implementing E2EE and federation in the future, but they're complex features, and I didn't want to end up like other community chat apps [3] that get criticised for broken core functionality and missing expected features while chasing those goals. I'm most confident on the backend and web app, so that's where I've focused. After some frustrating attempts with React Native, I'm sticking with a mobile PWA for now (including push notification support) while looking into Skip [4] for a true native app. If someone with more experience in native dev has any thoughts, let me know! Many tech-related communities that would benefit from not locking information into walled gardens still choose Discord or Slack over forum software because of the convenience these platforms bring, a choice that is often criticised [5][6][7]. I will not only work on adding forums and threads, but also enable opt-in publishing of forums to the open web, including RSS/Atom feeds, to give you the best of both worlds. ~~~ I don't intend to license any part of the software under anything but the AGPLv3, limit the number of messages [8], or have an SSO tax [9]. Business-oriented features like SSO will be prioritised on the roadmap with your support. You'd only pay for support and optionally for sponsored features or fixes you'd like prioritised. I don't currently plan on SaaS, but I'm open to support and maintenance contracts. ~~~ I want Fluxer to become an easy-to-deploy, fully FOSS Discord/Slack-like platform for companies, communities, and individuals who want to own their chat infrastructure, or who wish to support an independent and bootstrapped hosted alternative. But I need early adopters and financial support to keep working on it full-time. I'm also very interested in code contributors since this is a challenging project to work on solo. My email is hampus@fluxer.app. ~~~ There’s a lot more to be said; I’ll be around in the comments to answer questions and fix things quickly if you run into issues. Thank you, and wishing you all the best in the new year! [1] https://ift.tt/ke4X3wF [2] https://ift.tt/OEY3d17 [3] https://ift.tt/5309yvX [4] https://skip.tools/ [5] https://ift.tt/KsXB3Gr [6] https://ift.tt/dbsrfhC [7] https://ift.tt/m7GaezE [8] https://ift.tt/toSdqVZ [9] https://sso.tax/ https://fluxer.app January 3, 2026 at 01:30AM

Show HN: I mapped System Design concepts to AI Prompts to stop bad code https://ift.tt/6FAIgNp

Show HN: I mapped System Design concepts to AI Prompts to stop bad code https://ift.tt/B6ZEMWv January 3, 2026 at 12:15AM

Show HN: Feature detection exploration in Lidar DEMs via differential decomp https://ift.tt/JCI6pFA

Show HN: Feature detection exploration in Lidar DEMs via differential decomp I'm not a geospatial expert — I work in AI/ML. This started when I was exploring LiDAR data with agentic assitince and noticed that different signal decomposition methods revealed different terrain features. The core idea: if you systematically combine decomposition methods (Gaussian, bilateral, wavelet, morphological, etc.) with different upsampling techniques, each combination has characteristic "failure modes" that selectively preserve or eliminate certain features. The differences between outputs become feature-specific filters. The framework tests 25 decomposition × 19 upsampling methods across parameter ranges — about 40,000 combinations total. The visualization grid makes it easy to compare which methods work for what. Built in Cursor with Opus 4.5, NumPy, SciPy, scikit-image, PyWavelets, and OpenCV. Apache 2.0 licensed. I'd appreciate feedback from anyone who actually works with elevation data. What am I missing? What's obvious to practitioners that I wouldn't know? https://ift.tt/ND693KP January 1, 2026 at 05:59AM

Show HN: A Prompt-Injection Firewall for AI Agents and RAG Pipelines https://ift.tt/r6h0gHe

Show HN: A Prompt-Injection Firewall for AI Agents and RAG Pipelines We built SafeBrowse — an open-source prompt-injection firewall for AI systems. Instead of relying on better prompts, SafeBrowse enforces a hard security boundary between untrusted web content and LLMs. It blocks hidden instructions, policy violations, and poisoned data before the AI ever sees it. Features: • Prompt injection detection (50+ patterns) • Policy engine (login/payment blocking) • Fail-closed by design • Audit logs & request IDs • Python SDK (sync + async) • RAG sanitization PyPI: pip install safebrowse Looking for feedback from AI infra, security, and agent builders. January 1, 2026 at 02:31AM