04.02.2026 20:20
👍 3
🔁 1
💬 0
📌 0
I'm teaming up with @sa.muelcolvin.com to launch a new meetup series, Py AI.
Our first one is November 11 in San Francisco with talks from @pydantic.dev, @modal-labs.bsky.social, ReductoAI and @fastmcp.bsky.social.
Hosted by our friends at pebblebed.com
RSVP here: luma.com/gjtfecc7
Nah the client will listen for notifications from the server. You can notify it that tools have changed, you can use the client’s LLM so you don’t have to host one server side, and ask the user to fill out a form with structured data for a variety of use cases
Solves the identity state but, but doesn’t elegantly solve the bidirectional. If you slap bidirectional stuff onto hypermedia we’re back at inventing a new protocol anyway.
Thanks to @chris.blue for reading and giving feedback to an early draft of this!
In a recent post, @chris.blue argues that MCP is an unnecessary and temporary kludge on the decline and facing an uncertain future.
IMO it standardizes how agents have stateful, identity-aware conversations.
aaazzam.substack.com/p/the-apolog...
We didn't invent assets, but we're Prefecting them.
We spent a long time getting this right, and it paid off. Our goal was to let people show up with functioning python scripts and let them add data lineage with zero migrations, rewrites, cruft or goofy abstractions.
Wait so mcp auth is oauth through Claude AI, bearer header if by openAI SDK, and “secret url” if Claude SDK?
start of an era @jlowin.dev and @cicdw.dev
Great piece by @cicdw.dev on Task Runners in @prefect.io.
By decoupling the definition of what to execute from how to execute it, graduating from threads to @raydistributed is a line of code.
Who you callin slop
more like call of duty nazi zombies
when I see a /dags folder
just generate an llm.txt.py.pdf
I asked Operator to browse the Airflow docs and it just started googling “how to fake your own death”.
Wow I did not see that coming
@alternatebuild.dev watching me try to get pre-commit to run
😂 wut
Super excited for this one.
Prefect 💙 scitkit-learn
Yeah, it doesn’t hold a candle to how clean your implementation is here
In the Marvin codebase circa June 2023 there was AIModelFactory that did this that I don’t think ever made it to the docs (surprise surprise 😂).
Can’t wait until they ascend to “dropped out with a masters and went intro industry” level of intelligence.
I’ll allow it
Help!
I have a new favorite meme format but can’t decide on the best punchline.
🙇 #databs plz I’m begging you.
