A better permissions layer for coding agents. The tool works like auto-mode for Claude Code, so you can stay in the flow and only get prompted to allow or deny tool calls when it truly matters, but it is fully deterministic. My benchmarks surfaced that most Bash calls don’t need an LLM to be classified as safe, ambiguous, or dangerous. A deterministic classifier can auto-allow or block 95% of Bash tool calls as safe or dangerous, with only the remaining 5% being truly ambiguous or unknown.

Conclusion is permission reviews with LLMs like Claude’s auto mode or Codex auto review are like using a data center to flip a light switch - overkill.

The main benefit is that your agent’s autonomy can be governed deterministically through policies that can be stored at the user and repo level. The bonus is that you save tokens vs using auto modes.

https://nah.build

Diagnostic tool for all vehicles.

Describe your symptoms in as much detail as detail as you like and get a full diagnostic report with parts links, tutorials, price estimates, guides for diy

Includes car sales tool, generates all the documents you need for the dmv in your county/state

iPhone all connects to Bluetooth obd2 sensors for check engine lights and live driving data

Setup search alerts for “dream cars”

This one has been a blast to build

https://crewchief.cc

https://apps.apple.com/us/app/crewchief-auto/id6760673109

I am trying to build a simulation that lets a simulated organism come up with its own small language, purely learned from sensory input: https://github.com/JoergStrebel/VirtualZoo/blob/main/compute... I would like to implement the ideas put forward by Stevan Harnad in his symbol grounding problem paper (Harnad, 1990).

I've been learning to crochet. I'm trying to do more hobbies with my hands, but it's also pretty interesting from a mathematical perspective. The fundamental primitive (the chain stitch) is like a series of slip knots, and each stitch is reversible. So the piece is actually a series of reversible transformations. The yarn is sewn in at the end to secure it.

This has some interesting implications. If you make a mistake, you can always backtrack and try again. If you have a crocheted piece, at least in principle you could find the lose end, free it, and work back stitch by stitch to reverse engineer it. (In practice people don't seem to do a stitch-for-stitch reverse engineering just like you probably wouldn't bother reimplementing something line by line without a compelling reason, you figure out what's going on in the challenging places just by look and feel and improvise from there.)

I'm oversimplifying somewhat and there are some forms of crochet that include irreversible stitches, yarn can be felted together (entangled, like a cotton ball) to create irreversible bonds between adjacent strands, and often several panels/pieces are joined together irreversibly to create a larger piece.

Trying launch my Clover/Website builder integration.

Launching a niche RSVP system

A Menu builder

A context/domain aware private message service.

I am working on SpreadsheetPreview.com, a subscription service that gives you PNG previews of your uploaded spreadsheet (xlxs files).

On the server it opens a headless browser, where it converts the XLXS format to OGF (Open Grid Format), which is then rendered by DataGrid Toolkit, the engine behind DataGridXL v4. It then takes a screenshot of this render and sends it back to the requester.

Try out a few renders at https://spreadsheetpreview.com

I made a Python tool to build distroless container images for projects managed by uv. It draws inspiration from Ko from the Go ecosystem and works with/depends on uv from the Python ecosystem, so I smashed them together and called it Kuvo: https://github.com/hxtk/kuvo

It’s a hobby project in a very early state where it technically works but it’s missing several things I think it needs before I’d use it for anything serious. As of right now it isn’t even complete enough to dogfood a minimal container for itself without an intermediate base image because it can’t target a platform compatible with the distroless uv container image.

I'm working on [Context Bonsai][1] - LLM harness tools that allow the LLM to prune messages out of the context, leaving behind a summary and keywords instead. In addition to a "prune" tool, there is a "retrieve" tool that allows it to recall the messages if needed.

In addition to these tools, I'm also building automation that will port the tools from the reference implementation (OpenCode) to other harnesses (Claude Code, Cline, Pi, Gemini, Kilo, Codex, others to come?). As well as automation that will either cherry-pick or re-implement commits onto the latest head from upstream.

[1]: https://github.com/Vibecodelicious/context-bonsai-agents#con...

[2]: https://blog.vibecodelicio.us/posts/how-i-fixed-context-wind...

With AI IDEs, Personally I had to generate a tons of md files for planning a task, analysis on the code for something or other, a task doc for a feature - a summary to paste into the clickup ... and I saw many devs keeps on generating them, and all tucked away into folders. Good, okay.

For the company I'm currently working I had made a VSCode extension where I can sync the task doc with clickup via frontmatter.

I decided to take it to next level as a side project. I built a CI integrated, git-native, agent template transformable syncing pipeline with git MD files to any project management tools. That means, either you can save your md files vanilla in your wiki (thus using the clickup AI search to dig up later, get insights etc) or you can use a AI agent template transformer to turn it into a task template (Background, acceptance criteria, functional requirements etc.) and update or create a task on a board.

I've been working on it now. I don't know how it will fare, but I feel like product is coming up nice.

https://mdspec.dev

I'm working on a general repo shape/structure linter (language agnostic)[0] - the idea is to enforce things like directory structure, existence of various files (LICENCE, etc.), file naming patterns, jsonpath + schema over json/yaml/toml, absence of potentially malicious unicode. It comes with rule bundles for various languages/presets which can be combined and extended. A goal is for it to be very fast, and useable on huge monorepos. I noticed myself having to add various forms of validation/scripts when coding using AI, and decided to build a reusable, fast tool for this purpose instead of rolling validation scripts for each project.

[0] https://github.com/asamarts/alint