Underpinning my current app is an e2ee local-first sync engine, basically it is a traditional client-server sync (encrypted logs + snapshots sequenced with integers). It sends bundles of Loro CRDT operations. I wrapped the client side in WASM to power the web app and the CLI and have started a swift wrapper to port to native iOS. Bundle size is 3MB/1.2MB g-zipped so pretty happy with it. I've realised that web encryption is kind of bs (at least not as "WE CAN NEVER ACCESS YOUR DATA" as some vendors state) if someone else is distributing the app.
Over the last week I have done a lot of performance work & data remodeling - CRDTs are interesting because you can let data fall through the gaps if you're not careful.
The host made an offhand mention that there's probably a bunch of other similar sites that could be created with all the of useful but difficult-to-access government data out there. That sounded interesting, so I thought I'd give it a whirl!
Working on a few of them, including The Waterline (https://the-waterline.com/) for water info for the western US, The Scramble (https://the-scramble.com/) for egg prices, and The Dwell (https://the-dwell.com/) for container ship dwell times.
All pretty fascinating topics to learn about, plus it's been interesting to see how much of the website setup I can fully delegate to Claude. With Cloudflare to buy domains and put the sites up, a Google Service Account with access to Google Search Console and GA4 to create those properties and a Buttondown API key for weekly email sending, it's almost all hands off for me. Though it refuses to take control of the browser and create a new Buttondown account, which I was surprised is a red line.
- https://smacke.net/ffsubsync -- automagically synchronize subtitles, now purely client-side in your browser thanks to pyodide
- https://ipyflow.github.io/ipyflow/lab/index.html?path=demo.i... -- reactive python jupyter notebooks, again in the browser thanks to pyodide / jupyterlite
- https://smacke.net/pipescript/lab/index.html?path=demo.ipynb -- magritter-like pipe / placeholder syntax for ipython / jupyter, again able to run purely in the browser
- https://smacke.net/pycograd/lab/index.html?path=pycograd_sim... -- pyccolo and pipescript-powered autograd, once again able to run purely in the browser since numpy has a wasm target (notice a theme here :) )
I'm also working on an update to ShopifySharp, the .NET package I maintain for Shopify's graphql and rest APIs. I need to regenerate the graphql types and the fluent query builders for the July 2026 API version that was just released, and I'm planning on some extra QoL improvements that I've run into while using the package over the last couple of months. I particularly want to add some F# QoL features, since I wrote the package in C# but use F# in all my personal projects. (https://github.com/nozzlegear/shopifysharp)
Part of my job is to approve / reject MCP servers based on how secure they are and whether they are suitable for use in an enterprise environment. I was tired of my team being called the bottleneck to AI adoption, so I set out to automate the whole process.
I periodically collect the MCP servers and every new version from the Official MCP registry and assign them a score based on 29 distinct criteria like runtime guardrails (e.g. destructive tools, over broad permissions, rug pulls), SAST scans and transport & trust model.
As a result of this exercise, I found that 1 in every 10 MCP servers is pretty much unusable (score 40/100 or below). 18% of the popular MCP servers with 1000+ GitHub stars contain one or more security issues. 184 servers to date have changed their tool definitions after publication, which may indicate a "rug pull" attack.
I built this for security minded people who also want to be at the forefront of AI adoption and for security teams who are tired to be called the bottleneck.
Browsing the index is completely free, you only have to request an API key if you want automated, programmatic lookups for any workflow.
Feedback is always welcome!
StudyEngine is a webapp I'm using while doing my masters in comp sci. I upload lecture notes, textbooks, papers, etc. It then extracts topics and tracks my mastery of them over time. It uses an LLM to generate questions and flash cards. It loops in some newer learning science ideas. It tests recognition first(multiple choice), and then once a level of mastery is matched, it switches to recall. Working on adding RAG to it, so I can surface where in the source material something can be reviewed when going over quiz results. Currently just for me an some friends. If can get a good eval set up, I might work on optimizing cost and seeing if it could be opened up.
NomNominees is simple webapp that tracks James Beard, Great American Beer Festival, Festival of Barrel Aged Beers, and other awards. I use it when I'm traveling to find places to check out. Even just a cluster on a map shows me neighborhoods I might want to check out.
It helps me to automatically save a tab that's not been used in a while so it auto-closes it but saves it as well as having the ability to snooze a tab like how you'd do it in gmail.
Everything is locally stored with 100% privacy in mind.
The core idea was that I've always been a lousy notetaker, even going back to my school days years ago. I'm great at one-off and one-liner notes and occasionally more in-depth notes, but tend to not flesh them out fully enough to make them worth re-visiting.
This has been a struggle even as an engineer sitting in meetings or trying to absorb new information when starting a new job and ramping up.
Logbook is meant to use an interaction paradigm we as engineers are using very often these days: it's a terminal UI in the vein of Claude Code, OpenAI Codex, etc.
It's targeted at the entry of free-flowing thoughts but you can also write longer notes by launching your default shell editor from within the tool.
Each note is saved as markdown with some metadata and that metadata is then saved to a local SQLite DB.
For the LLM side, the tool extracts useful metadata from those notes and then performs some local ranking/categorization. It then has the ability to send a note or some metadata to a provider of your choosing (it's straightforward to use OpenAI or something more broad and customizable like OpenRouter) for further enrichment or filtering.
A couple examples of the currently implemented slash-commands: `/related` can be used to find related notes; say you've been scribbling down notes about OAuth or MCP servers and want to gather up the most relevant notes to one of those topics. Or you can use a `/gaps` command that'll help you find things you've taken notes about but without properly defining or providing context around them (i.e. you mention ID-JAG for OAuth but never actually say what ID-JAG is, this command will tell you this so you have a chance to review what you previously wrote and can then define exactly what that keyword is about).
It's still very much a work in progress. It's not meant to be a full-fledged note-taking app a la Obsidian or anything like that. I've just always preferred taking notes in markdown or plain text and this is a great way to continue doing that while also making enrichment of the notes pretty simple.
You may ask "why not just use agent memories?" I don't really like the idea of tightly coupling notes with codebases or agents and I don't find the current UX very intuitive at least for the way I prefer to take notes.
Ask the same engine the same question twice and you get different answers, different citations, sometimes a different opinion of your brand, so figuring out how best to present this has been a fun product problem to solve.
It also tries not to be yet another dashboard: instead of just analytics, an agent turns the findings into a ranked list of "ship this fix" todo items.
Was using this only for my self, but i think it might be interesting for other people as well.
It's for iPhone, and for the best experience, Apple Watch. It's very early, playable via TestFlight, and I would love feedback! There's a TestFlight link at: https://reverdure.yourstrategy.co
Sharable, real-time synced maps, Google Docs for maps basically.
I think the coolest part is the import feature where you can paste a link to a video or article and it pulls out places and enriches them with images and a description. You can also write your own notes, vote on places to go with friends, and apply colors. Right now I am working on user acquisition and experimenting with different marketing approaches.
If you're an early stage b2b founder, I'd love to hear your feedback about TractionBeast.
[1] https://hyperclast.com/ - fast, self-organizing, self-hostable replacement for Notion
Aggregator for new posts in build threads from 277 old-school DIY forums.
Build threads of people building cars, 4x4s, motorcycles, boats, airplanes, hot rods, musical instruments, etc.
Good luck with your build and perhaps I might get interested in future too as I did once have a thought that having a custom car to me would reflect more cool-ness than an expensive one. I am really interested by small cars, perhaps retro. I imagine my favourite car to be somewhat like the car that Ryan gosling drives in La La Land.
but a cool project nonetheless, certainly thinking about it inspires a bit of car enthusiasm within me even though I am not that much of a car fan so much right now so a really cool project if it can help more people feel this spirit. good luck :-D
I have a question but how does building new (retro-inspired?) cars go about in terms of pricing. I feel like they might be too costly to get custom-built and that If I really ever in my life go about doing this, I would prefer DIY but I still imagine that it might be too expensive or hard to make a car. Are there any go-to cars which are easy/recommended within this space and how does it compare off economically and what are the technical expertise that you require with this type of stuff?
Once again, I wish ya good luck in the project and would love to hear your answers for some of the questions I have!
You're right that getting a car custom-built is where the costs add up quickly; easily north of $50K. Most of the cost is labor, which is $0 if you do it yourself. Some of the projects are much easier than others. If you want to fall down a rabbit hole, look in the kit car and hot rod categories; lots of affordable and small builds in there. The Buick Riviera in La La Land is more of a resto-mod cruiser project, but the small/retro itch is exactly what the kit car category scratches. The first step is to find a forum where people are building the car you like, and start following related build threads. That's the majority of my social media intake these days.
https://github.com/thegagne/aep-conformance-test
Did pretty well, only took a day or so. I first had it inventory every MUST, SHOULD, and MAY in the spec, and then let it rip. I did guide it quite a bit to get what I wanted, but at the end I’m pretty happy with it as a first draft.
Helped me learn the spec and will be helpful to hone my dotnet AEP server, and aepbase.
There already existed an aep e2e validator which does a similar thing, but this is more thorough and generates a nice report. It will tell you not just whether your API follows the spec, but also what parts of the spec it does not implement.
OrcaBot was my Jan+Feb attempt to defeat the lethal trifecta whilst offering all the bells and whistles of a claw like sandbox: https://orcabot.com/blog#breaking-the-lethal-trifecta
This month I've been working on the free desktop version which is available as of today but probably carries a few too many bugs to not be worth promoting just yet.
It came from a frustration that I needed to switch between the browser and the IDE to navigate through the code and leaving comments on Gitlab at the company.
So I thought it could useful to create something and let it be accessible to the public as open source.
link: https://github.com/LuyandaLia/reviewflow
In a nutshell, it accepts draft comments, which can be modified and submitted.
It auto configs the env for Python as it uses FastAPI for calls to Gitlab.
It's my initial attempt. Suggestions, reviews, contributions are invited.
One love
It's been fun dealing with memory and C's weird design in this age of agentic coding.
https://github.com/keloran/tiny-dfr
Unfortunately due to the way GitHub defaults to creating prs in the parent fork, I have accidentally created a few invalid prs in asahi before I was ready, and now am banned from creating a good upstream one
We just launched a couple weeks ago and we’d love any feedback or suggestions!
Search is currently provided by the Radio Browser API, but I'm now building my own station API with proper metadata and thumbnail coverage. A station discovery page with most played stations is also in the making.
I see a lot of new (and, to be frank, a lot of mature ones) HR tools are just wrapping Chatgpt around resumes (almost like "OK, now match this resume against this job posting and tell me if applicant fits"), which introduces a massive bias/inference problem.
I decided to build the exact opposite – a deterministic, math-driven fitness engine. It extracts structured scorecards from both CVs and job requirements and mathematically matches them, so you can actually review the exact reasoning behind why a candidate scored a, say, 85%. This fitness value is specified at every interview step – as applicant goes through an interview process their scorecard is updated at all steps.
If anyone here builds in the HR space, I’d love your feedback.
Open to feedback and missing pieces.
Realize that I'm really bad at marketing. Trying to work on it.
It lets you take a picture of video games and shows price comparisons for the major buy lists.
After some time I figured the best use of AI is to produce even more AI-related slop and spend my occasional 2 dollars on the deep seek model to do it.
Models are fun when given a stable identity and made aware of it.
It uses DuckDB to expose a sql query interface in the website itself because I wanted to give the freedom to just do something interesting with the data.
My friend John had an idea which I really liked so I added "john mode" which shows what he was suggesting :-D
I think that Hackernews might like it but honestly, I have probably just made it out for myself and also as something to just share casually with folks on hackernews and other websites and hopefully I am able to help people and myself in some way with this website.
Open to some feedback as usual (for mostly all my projects really) and thanks for reading and have a good day dear reader and hey perhaps give my website a try!
www.memoryplugin.com
I know what motivates me: seeing progress. The feedback loop of "do X, see Y gain" is what keeps me going.
So I started building an integrated dashboard that can aggregate data from multiple systems:
- My digital scale
- Apple Watch (sleep + running performance)
- Beastmaker Motherboard, which is an electronic board that you attach a hangboard to and it shows you various stats like how much force you're applying
The idea is that every morning I'll open the dashboard and be able to see exactly how much progress I've made the previous day: weight loss, strength gain, cardio performance.
It's an interesting problem. There's essentially two parts to it: Apple Health, which aggregates data from the scale and the Apple Watch and can POST-export it hourly, and the electronic board, which sends data via BLE in real time. The destination for both of these will probably be an always-on Raspberry Pi 5, but I haven't decided yet. Then I'll have a small server app that can pull the data from the Pi and draw some fancy charts.
The idea is to see trends and try to apply AI for correlating, at the first glance, completely unrelated data layers. Example how I'm thinking about this one: there's somewhat clear correlation that I sleep better when I do above average steps per day. How is my sleep quality affected if, let's say, I did above avg steps with a bad air quality at that time? (i.e. wild fires / pollen season / etc.)
I've built a Go application to ingest those data sources and currently finishing my first import use case - Apple Watch data.
Would be happy to connect and chat about this.
However, LLM coding has made coding less rewarding so… Im thinking about starting a new hobby as coding for fun has become prompting.