Kyle Kelley

I love when Claude says it's going to take 2 to 4 hours for something that I know will take it like less than 30 minutes in a code base.

Generative AI has the memory of a goldfish and a million eyes.

I need some way for my agents that operate on behalf of me on GitHub to indicate as such. Aside running and orchestrating these like ZERO places are making it possible to provide attribution and authentication for agents in a clean way.

Errors as product placement

upstream connect error or disconnect/reset before headers. retried and the latest reset reason: remote connection failure, transport failure reason: delayed connect error: Connection refuse

Speed of development wholly dependent on how fast testing and especially end to end testing can be done.

Split the Kernel by rgbkrk · Pull Request #6 · ad-si/Woxi Few new things here that should make it easier to hack on adding to the kernel. The read write pairs from the shell (and control) sockets can be split() so you don't have to manage ownership an...

annnd the final one for now until I later go dig into your implementation. github.com/ad-si/Woxi/p...

Bring Woxi up to date with jupyter crates by rgbkrk · Pull Request #5 · ad-si/Woxi Hello from BlueSky! Runtimelib, jupyter-protocol, and zeromq maintainer here. Delighted to find your kernel so I can do some symbolic maths for funsies. I was looking around to see who was using ru...

First PR, as promised. github.com/ad-si/Woxi/p...

runtimelib/jupyter-protocol maintainer here, I'm going to send you some PRs to bring you up to date and also clean up a few things. I think you'll enjoy having a `.split()` shell too.

Commisars hating the new regime

GitHub - ad-si/Woxi: Wolfram Language / Mathematica reimplementation in Rust (Wolfram oxidized) Wolfram Language / Mathematica reimplementation in Rust (Wolfram oxidized) - ad-si/Woxi

Wow. @adriansieber.bsky.social (re)implemented the Wolfram language in Rust github.com/ad-si/Woxi

and exposed it as a jupyter kernel. That's crazy.

I don't see any documentation

You would not believe the things I can now blame on AI.

Mmmmm scrolly terminal in @zed.dev's REPL output (@jupyter.org backed). PR: github.com/zed-industri...

Whoops! I posted the wrong image. This one has all of them as implemented today.

Nice, I can use that!

@prefix.dev 👀

Adding little icons for the environment type to the new desktop notebook app.

Display updates mode matplotlib Display updates mode matplotlib. GitHub Gist: instantly share code, notes, and snippets.

gist.github.com/rgbkrk/d7450...

Yeah I'll hand y'all the script via gist later tonight or tomorrow.

Custom keymaps in the zed editor. repl:run is triggered by keystrokes "x j", "x o", when Editor && VImControl is on.

At least for me I have a custom keymap so that in vim mode I use `x o` for run and move down.

Nice that was pretty fluid to run through. Not sure why my plot doesn't work though that might be because I'm using `display` with the update display protocol from Jupyter.

repl: Add treesitter block detection to Run action by rgbkrk · Pull Request #49636 · zed-industries/zed When the cursor has no selection, the Run action now uses treesitter to detect the enclosing top-level block (function, class, loop, etc.) and sends it to the REPL. This lets people send whole bloc...

PR: github.com/zed-industri...

Making the @jupyter.org backed REPL in @zed.dev be more language aware via treesitter. Run blocks of code. No need to select segments or denote them as cells.

That's even better, directly integrated at the import. For Jupyter that needs some new messages. I miss when we were pushing the protocol forward.

Exclusive: Pentagon threatens to cut off Anthropic in AI safeguards dispute Anthropic has not agreed to the Pentagon's terms and defense officials are getting fed up after months of difficult negotiations.

Much respect to @anthropic.com for holding to standards. www.axios.com/2026/02/15/c...

Screenshot of a desktop notebook app with a modal titled “Review Dependencies.” It lists PyPI packages pandas and numpy and asks the user to “Trust & Install” or “Don’t Install” before running code. The notebook UI behind it shows a dependency panel and Python cells importing those packages.

Screenshot of a desktop notebook app showing a “Review Dependencies” modal for Conda packages from the conda-forge channel. It lists numpy, scipy, matplotlib, and requestz, with a warning that requestz is similar to “requests.” A highlighted notice says “Potential typosquatting detected” and asks the user to verify before approving. Buttons read “Don’t Install” and “Trust & Install.”

What if notebooks let you install dependencies directly and when loading someone else's, they inform you of what's going to be set up in the environment?

Yes this is a new desktop app.

I've been in the AI world for like two years now and I am continuously amazed by how many problems are no longer technical problems.

The bigger cost of implementing automation is fine and institutional inertia.

Backlogs can only grow.

Kind of fun to use inside of Zed.