Hi HN,

I built a small QR code editor after repeatedly running into the same problem: once a QR code is printed, the destination is frozen.

This tool supports both:

Static QR codes (generated entirely client-side)

Dynamic QR codes (editable destination after creation)

Dynamic QR codes come with a minimal dashboard where you can:

Update the QR destination without reprinting

See basic scan stats (count, device, time, approximate origin)

I intentionally kept the feature set small and avoided heavy analytics or tracking. The goal is something simple that works well for menus, flyers, events, and small businesses.

The app is live here: https://qrcodeeditor.com/

I’d appreciate feedback on:

What’s missing for this to be useful day-to-day

Whether dynamic QR codes are something you actually need

What you’d expect before paying for a tool like this

Thanks for reading.

I run a newsletter tracking NYC startup jobs and built this Airtable database as a resource. Updated weekly with venture-backed companies hiring across all functions. Sharing in case it's helpful to the community.

I've worn an Oura ring since 2020 and I've been that guy who cross references HRV with everything from supplements, late meals to quitting nicotine. I use the data to tell the difference between 'I feel off' and 'my resting heart rate is 8bpm higher than baseline'. And also for introspection, and to make better decisions in the future as Claude has access to 6 years of journal entries as well. (Yep privacy concern, but I find it worth it for me)

What bugged (lol) me is that I was spending too much time exporting data and manually correlating things. Earlier this year I tried feeding 5 years of data into a custom GPT, and the responses were actually (sort of) useful if it didn't choke.

So I built this as an MCP server, I had no idea how MCPs functioned so this was a good way to learn. It fetches fresh data on-demand rather than choking on a static dump, and the statistical analysis (correlation, outlier detection, trends) happens server-side before the LLM even sees it. When I ask "what predicts my best sleep?" it actually computes the answer instead of pattern-matching on vibes.

Works with Claude Desktop. Developed with Claude, see CLAUDE.MD TypeScript, ~600 tests, MIT licensed.

https://github.com/mitchhankins01/oura-ring-mcp

- Added support for document metadata preservation - Bring your own LLM - No OS limitations

The Project: I am building R3-Engine, a from-scratch, local AI inference engine for Microsoft's bitnet-b1.58-2B-4T. It is written in 100% Safe Rust, natively cross-compiles to Wasm SIMD128, and uses Zero heap allocations in the execution loop.

The Physics: By mapping a 64-byte aligned .r3 file directly from NVMe to CPU L3 Cache (Zero-Copy) and using AVX-512 VPOPCNTDQ for branchless math, the Ryzen 9950X3D achieves 117 Tokens/Second latency.

The Problem: The AI is mute (Outputting <unk>*)* The matrix multiplication pipeline is mathematically complete, but the output is stuck at Token ID 0 (<unk>). The issue lies in the transition between the quantized weights and the float-based non-linear activations.

Where I need expert input:

    Weight Tying in BitNet: Microsoft's 2B model ties Embeddings with the LM Head. I am cloning the embedding matrix for the output projection, but I suspect a scaling factor is missing.

    RMSNorm & SiLU in 1.58-bit: How should the raw integer accumulators (from the VPOPCNTDQ loop) be scaled before entering the SiLU activation and the subsequent layer?

GitHub Repo: https://github.com/r3-engine/r3-engine

If you know the physics of LLM Logit Sampling or ternary activation math, I would love your eyes on the codebase.

Hi HN,

This is Bookface — a TypeScript wrapper for Meta's developer services, born from frustration at Meta's lack of easy developer integration.

The current API libraries you'll find for Meta's products are either very limited (and use hacky browser simulation) or unacceptably old/unmaintained. This is a library that aims to:

Interact with Meta's developer services the way Meta intended, although with way more ease.

Support all of Meta's platforms (e.g., Facebook, Instagram, Threads, WhatsApp, etc.).

Be consistently updated with Meta's latest API changes.

Overall, make things faster for people trying to use Meta services.

For background, I've been working on this library on and off for about a year and a half, slowly adding features as I needed them. I originally created this library for another project when I was seventeen. Two years later, I have been putting off release for too long, and this library desperately needs other experienced hands.

The reason for this post is to get some urgent feedback on what syntax feels good and what doesn't; to test that what's currently added is working as intended (I am the only tester); and, crucially, to find other maintainers. I want this to be a long-term project, and maintainers would help smooth out the rough edges of my mistakes and hopefully keep this project going for a while.

I should note that this is sort of a preliminary release post since this library is still in its very early stages (lacking features), and I'm really just looking to get things into motion. Please check it out!

https://facebook-js-sdk.vercel.app/

(The current, early-stage website. I’m working on getting a better domain.) (Also forgive the name discrepancy with the website and repo — the name is still up in the air and I'm open to suggestions.