Overlay of the logo for nave on top of the output from the discover entrypoint running with a --prune flag (removing forks from the already cached repos)

we rollin

Prompting in one window and writing a blog simultaneously in another is NUTS

Rust API 👀 cuenv.dev/reference/ru...
++ github.com/cue-lang/cue...

CUE - Configure, Unify, Execute CUE makes it easy to validate data, write schemas, and ensure configurations align with policies. CUE works with a wide range of tools and formats that you're already using such as Go, JSON, YAML, TOML, XML, OpenAPI, Protobuf, and JSON Schema.

CUE: "Configure, Unify, Execute" 🤔 github.com/cue-lang/cue

A data constraint language

Some motivating thoughts on repo fleet-level ops cog.spin.systems/fleet-ops

(´-`) .｡oO( a fleet-level repo control plane would fix me… )

Separate CI test legs for --frozen and --upgrade --resolution lowest-direct

MCP Python SDK does this github.com/modelcontext...

When publishing libraries, it is recommended to separately run tests with --resolution lowest or --resolution lowest-direct in continuous integration to ensure compatibility with the declared lower bounds.

Apparently this should go in CI as a separate matrix leg docs.astral.sh/uv/concepts/...

Need to learn the ways of `uv sync --resolution=lowest-direct`

GitHub - cargo-public-api/cargo-public-api: List and diff the public API of Rust library crates between releases and commits. Detect breaking API changes and semver violations via CI or a CLI. List and diff the public API of Rust library crates between releases and commits. Detect breaking API changes and semver violations via CI or a CLI. - cargo-public-api/cargo-public-api

Nice: `cargo public-api diff` github.com/cargo-public...

There’s definitely an argument that GitHub Notifications inbox-like design (a single feed with filters) has all the drawbacks of an email inbox, making things easier to miss

User reports, dependency update bots, long-running discussions, (…) all go in one undifferentiated mass

Woman filming out the bus window through Trafalgar Square talking to herself constantly in a hushed tone, can briefly make out "A.I.'s just told me...", incessant feed of comments coming into her stream

Landan bruv!

£11 central pints observed in the wild 🫣

Charts with absurdly fast results, a new embeddings model from Flower Computer

Comparison of fast static embeddings vs MiniLM-L6 and other models

Scenes www.flowercomputer.com/news/fast-st...

Where AI Can Be Useful Everything I've said so far has been about Al producing code, and the same applies to text. There is a very good reason for me to write this with my own hand and brain; it is my excuse to think things through! However, that bug that Claude tracked down was clearly a case where Claude was faster than I would have been, and the debugging process itself would probably not have contributed much to my understanding of the project. I am much more open to using Al for bug finding than for producing code, perhaps I will experiment more with it. At the end of writing my thesis, I ran the document through Microsoft copilot several times, asking it to identify obvious issues with the language. It worked! It found a bunch of typos, grammatical errors, awkward wording etc. that I would have missed even if I had time to read through the 200 page thesis 10 times in the final weeks before the deadline. My prompt for this task was deliberate however, "point out issues", not "fix the issues". I used it as a tool to scan through text to find issues, not to reformulate anything. I have been considering experimenting with using Al to notify me of outdated documentation for Spade. We regularly have confused users who read some documentation and are surprised when the language has changed. I'm not reading through the docs on every release, so things like that do slip through, even though we now have some systems in place to prevent massive discrepancies. But anAl could also help here, we could give it a changelog, and it could suggest places in the documentation where things might need changing. All of these things fall into one category of problems, problems where something has to go through large chunks of code or text to identify something, but where once it has been identified, it is easy to verify whether the answer is correct. This is a category where I could see myself using Al, perhaps today, certainly in a potential future where we have good open source LLM

Suggests LLMs would be better suited to quality inspection tasks (debugging code, reviewing writing for grammar etc, or docs for staleness), pushes back against exceptions for throwaway code as those projects can also provide mastery experiences that benefit more serious projects

Well presented reasons for not using AI in software dev

The no. 4 bus to Cambridge front sign has text above it saying "this bus is a safe space"... new one

If you have a dynamic site you can add a POST endpoint to append to a ‘decisions’ TOML that’s also cached in localStorage, or GH API, etc.

Obvious next step tho is to also automate LLM triage from that, to give some reasonable starting point (expect the well modelled inputs would make it effective)

The free text can be HTML, YAML, etc — I call the things I expect to find in there “signals”, and transform them into well-typed Pydantic models scoped to a domain stored as TOML (separation of concerns: one module makes one TOML) that are then aggregated into a parquet/exported to JSON for the site

Pattern of operations I’ve been doing a lot lately = using LLMs to transform natural language description of the genres of info to expect within free text, thereby producing software encoded rule-based parsers (keyword/regex etc) that produce featurised datasets I view in a site & trivially annotate

Incredible things are happening in 127.0.0.1:8000

2026 shaping up to be all about bespoke triage systems

Interesting, do you have samples of data labelled this way shared online anywhere?

Nodes, Edges, and the Historian’s Craft: Knowledge Graphs as Research Notes How moving from Word docs to network graphs transformed my research workflow and how LLMs made it possible.

Knowledge graphs are transforming my research methodology. Claude Code allows me to build knowledge graphs as I take notes from primary sources.

open.substack.com/pub/computat...

🧘‍♂️

Listen to PyPI - Hear the Python Package Index in real-time Every PyPI package release becomes a sound and a glowing circle. An ambient soundscape driven by real-time Python community activity, by miketheman.

I made a thing, a soundscape based on #PyPI package data feed updates 🎶🐍📦🎶

Maybe you'll enjoy it too?
miketheman.github.io/listen-to-py...

Just made an acquisition… I’m data rich

Setting up mitmproxy SOCKSv5 proxy on Firefox obscure technical resolutions re: errors, installation quirks, custom setups etc. - lmmx/devnotes

Wrote a note-to-self how to set up mitmproxy (SOCKSv5 proxy on Firefox) to intercept your own web traffic github.com/lmmx/devnote...

Many sites serve JSON from internal APIs that turn into HTML in the browser, this lets you easily store those locally to do whatever you please with

“zappa: an AI powered mitmproxy” geohot.github.io//blog/jekyll...

> Download mitmproxy and configure Firefox to use a SOCKS5 proxy and install the required cert to proxy HTTPS traffic. Write a plugin for mitmproxy to route all website traffic through Qwen … to remove all ads, popups, …