trevor manz

trevor manz

@manzt.sh

middle child, creator of anywidget.dev building @marimo.io; prev @harvard.edu

529 followers 578 following253 posts

Building bridges between open-source projects is a big part of why I joined @marimo.io. I got to know the @prefix.dev folks through @scipyconf.bsky.social, and worked together to make computational research easier to reproduce and share. Special thanks to @julianhofer.eu and Ruben Arts!

marimo@marimo.io

marimo now supports @prefix_dev Pixi sandboxes! Share portable notebooks with dependencies beyond Python, including CUDA-enabled @PyTorch, ffmpeg and R. Already using Conda or Pixi? Bring your ecosystem to marimo’s reactive notebooks, with Conda and PyPI requirements saved alongside your code.

Scientific work often spans languages and tools. Pixi draws on Conda to bring in more than Python packages. For example, this notebook uses @marimo.io reactivity to filter a table interactively in R:

Reworking our sandbox model brought improvements for both Pixi and uv users. The notebook’s manifest now drives setup and package changes, keeping the environment in sync. A new UI lets you edit requirements, track progress, and retry failed setup without restarting marimo.

my cats kill all the bugs it’s awesome

pixi 0.76 can now run standalone python scripts in isolated, on-demand environments. dependencies live inline via PEP 723, like uv supports, but pixi can layer on conda and system dependencies too. fun to build this with @prefix.dev folks. coming to a @marimo.io notebook near you SOON.

github.com/prefix-dev/p...

github.comfeat: Support PEP 723 scripts with `--script` by manzt · Pull Request #6648 · prefix-dev/pixiThis PR follows the discussion in #6632 and reworks the earlier manzt/pep-723 prototype. Edited This PR originally added a pixi script command and conda-exec support. After review, it now uses --sc...

marimo.io/pair has a fresh landing page and some updates! for the launch, i wrote about why we moved beyond mcp and how we gave the agent @marimo.io itself as a tool through python.

marimo.iomarimo pair — a thinking canvas for you and your agentGive your agent a marimo notebook. The ease of chat, the rigor of software.

blog: marimo.io/blog/noteboo...

marimo.ioImplementing marimo pair: letting agents use notebooks as a toolWhy we stopped wrapping the notebook with MCP tools and let the agent work inside the kernel.

@marimo.io pair is now available in the Claude Codex and Codex CLI marketplaces, and today’s release of the @vscode.dev / cursor extension integrates directly with the native notebook editor. happy pairing

i’m @pycon.us with the rest of the @marimo.io team. come say hi!

got to talk through more of the engineering behind marimo-pair, writing skills, and interactive visualization. thanks for having me on! www.youtube.com/watch?v=K791...

youtube.comPair Programming with AI in Your Python Notebook — with Dr. Trevor ManzYouTube video by Super Data Science: ML & AI Podcast with Jon Krohn

Really enjoyed coming on Real Python and chatting about marimo-pair!

Christopher Bailey@digiglean.bsky.social

It was great having @manzt.sh on the show from @marimo.io! Thanks for sharing what you’ve been working on for data scientists and notebook users.

need to rewire my brain

on the orange website www.youtube.com/watch?v=6uaq...

trevor manz@manzt.sh

building marimo pair has been a new kind of engineering (for me). a lot changes when the consumer of your api is a model rather than another program. we’ve shipped “breaking” changes without upgrading or migrating the skill. the models just figure it out. hope you like it :)

building marimo pair has been a new kind of engineering (for me). a lot changes when the consumer of your api is a model rather than another program. we’ve shipped “breaking” changes without upgrading or migrating the skill. the models just figure it out. hope you like it :)

marimo@marimo.io

Introducing marimo pair: a skill that drops agents like Claude Code inside a running notebook, letting them read variables, run code, and manipulate UI elements. It also lets agents use marimo as a standalone REPL. youtube.com/watch?v=VKv... (1/n)

youtube.comWe gave Claude Access to All Python VariablesIntroducing marimo pair, a new way to have your agent interact with a live marimo notebook. Related blogpost:https://marimo.io/blog/marimo-pairGithub reposit...

the latest release of the @marimo.io @vscode.dev extension adds LSP diagnostics for marimo rules multiple variable definitions and dependency cycles are flagged as you type:

Introducing quiver: a (type-safe) place to keep your arrows @arrow.apache.org quiver is a small schema library for Arrow, built on @idl.uw.edu flechette - Define table schemas in TS - Validate schemas at IPC parse time - Get back fully typed Tables (no type-casting) github.com/manzt/quiver

View full thread

One thing I'm most happy with is the testing setup. Wrote a utility that snapshots the inferred TS types _inline_ alongside runtime checks of the actual values. Change something in type inference (quiver) or runtime (flechette) and the snapshots break.

I started quiver during my PhD but was stuck on the last mile — exhaustive tests that match inferred types to runtime behavior. The snapshot idea finally unblocked it and caught some edge cases and bugs. (Thanks for the help, Claude!). Give it a try and let me know what you think!

Introducing quiver: a (type-safe) place to keep your arrows @arrow.apache.org quiver is a small schema library for Arrow, built on @idl.uw.edu flechette - Define table schemas in TS - Validate schemas at IPC parse time - Get back fully typed Tables (no type-casting) github.com/manzt/quiver

flechette's `tableFromIPC` is to Arrow what `JSON.parse` is to JSON. quiver is like Zod or Pydantic for that — define your expected schema, check it at parse time, and get a flechette Table with the types overlaid

flechette columns have `toArray()` and `at()`. With quiver, you pick your level of strictness. The types propagate through `getChild`, `select`, `toArray`, iteration, even into nested structs and lists.

spent some time last weekend getting higlass to play nicely inside @marimo.io molab.marimo.io/notebooks/nb...

some of the first things new users notice and love in @marimo.io are our opinionated formatters (e.g., our table for dataframes) for v0.20.0 I worked on one aimed at AI engineers: a rich formatter for @pytorch.org `nn.Module`

always fun to jump back into some interaction design work some details: - collapsible tree → progressive disclosure of full config - color → module type (i.e., the `nn` submodule) - opacity → trainability (e.g., frozen or no params)

some of the first things new users notice and love in @marimo.io are our opinionated formatters (e.g., our table for dataframes) for v0.20.0 I worked on one aimed at AI engineers: a rich formatter for @pytorch.org `nn.Module`

always fun to jump back into some interaction design work. some design details: - collapsible tree → progressive disclosure of full config - color → module type (i.e., the `nn` submodule) - opacity → trainability (e.g., frozen or no params)

my new bug, minnow

pals

breaking news: dads are learning @marimo.io

thanks for having me!

Talk Python Podcast@talkpython.fm

Latest episode with @manzt.sh and host @mkennedy.codes is out: #530: anywidget: Jupyter Widgets made easy talkpython.fm/episodes/sho...

talkpython.fmtalkpython.fm

byoe (bring your own editor) for @marimo.io is here. built from scratch with LSP + uv marimo.io/blog/vscode

marimo.ioAnnouncing our VS Code and Cursor extensionBringing a native marimo experience to your favorite IDE

👀 preview: @marimo.io notebooks are Python first inside @vscode.dev. we're also integrating managed sandboxed environments for PEP 723 notebooks, powered by uv ofc. forget venvs

look ma, @marimo.io 🤝 cursor (w/ the upcoming @vscode.dev extension). things are starting to come together...

a new life for my 2019 macbook pro

marimo@marimo.io

We've all seen dataframe spaghetti code out there. But the remedy ... is pretty easy: just start using the pipe method! The moment you see the `.pipe()` in action, you'll never want to write pipelines without it. See the reason why here: www.youtube.com/watch?v=J0P...

youtube.comThe One Trick That Fixes Pandas Spaghetti CodeIn this video I am going to show you 6 reasons why the `.pipe()` method basically fixed all dataframe code out there. Python notebooks turn messy very quickl...

i’m used to hearing from strangers when my open-source "fails" them… this made my day

still lots to do... but we're cooking on a native @vscode.dev extension for @marimo.io notebooks

if interested in creating anywidgets of your own, our tutorial was finally shared to youtube: www.youtube.com/watch?v=frEo...

youtube.comAbdennur, Lekschas, & Manz - Bring your __repr__’s to life with anywidget | SciPy 2024YouTube video by SciPy

materials github.com/manzt/scipy-...

github.comGitHub - manzt/scipy-2024-anywidgetContribute to manzt/scipy-2024-anywidget development by creating an account on GitHub.

learning a new api? `mo.inspect()` your objects in @marimo.io. no more dir() / help()-ing around `<object at 0x...>`...

ok actually this time

trevor manz@manzt.sh

ok fine i’ll learn jj

Observable-like cell minimap coming soon to a @marimo.io notebook near you 🫡

github.com/marimo-team/...

github.comAdd reactive notebook minimap for navigating and visualizing dataflow by manzt · Pull Request #5707 · marimo-team/marimoThis PR introduces a reactive notebook minimap (similar in spirit to Observable), but adapted marimo where multiple variables can be declared across cells. Selecting a cell highlights its upstream ...

happy to finally share what we've been cooking 🍳

marimo@marimo.io

Announcing molab: a cloud-hosted marimo notebook workspace with link-based sharing. Experiment on AI, ML and data using the world’s best Python (and SQL!) notebook. Launching with examples from @huggingface, @weights_biases, and using @PyTorch marimo.io/blog/announ...

marimo.ioAnnouncing molabCloud-based notebooks for our community

trevor manz reposted

@nvictus.bsky.social

Replying to @nvictus.bsky.social

I’m also excited to be presenting Oxbow as part of my talk on composability at the #SciPy2025 Conference on Wednesday! Hope to see some of you there. cfp.scipy.org/scipy2025/ta...

cfp.scipy.orgBreaking the silo: composable bioinformatics through cross-disciplinary open standards SciPy 2025The practice of data science in genomics and computational biology is fraught with friction. This is in large part because bioinformatic tools tend to be tightly coupled to file input/output. As a res...

@marimo.io now has a tree-based UI for uv environments add/remove/upgrade deps the uv way — no more flat pip-style lists or accidentally pip-ing your env

spending some time on our uv integration in @marimo.io – starting with things like this selector for package “extras” what’s frustrated you about using packages in notebooks?

preparing for some potentially unhinged things...

have a nice budget to upgrade my at home setup. what’s one thing that ~actually~ made a difference in your space?