Define `pudl` Agent Skillset · Issue #2 · catalyst-cooperative/skills Overview This epic tracks potential improvements to the PUDL agent skill, which is meant for data users who are not necessarily working in or with the PUDL repo, and are accessing PUDL data and met...

Also working on an agent skill to help them navigate the hundreds of tables and thousands of columns effectively

github.com/catalyst-coo...

We'd love to be able to use a beautiful spatial overview like this to offer folks an entrypoint into the more detailed and less mappable datasets. Utility finances, generator operating costs, capacity factors, emissions control equipment, etc.

If you ever want to bring in more detailed US electricity system data, time series, etc. take a look at our CC-BY-4.0 Parquet outputs and get in touch!

catalystcoop-pudl.readthedocs.io/en/nightly/d...

The world's electricity infrastructure, mapped.

Tragic mistake... Anthropic leaks Claude’s source code YouTube video by Fireship

(Actually) open AI frameworks are chasing the frontier pretty hard. It seems unclear how different the underlying LLMs really are, and how much of their apparent differences are due to the tooling that surrounds them. Especially with, uh...

www.youtube.com/watch?v=mBHR...

Deep Agents: LangChain's SDK for Agents That Plan and Delegate When you type a question into ChatGPT, the model only has what you typed to work with. But tools like Claude Code can plan, iterate, test, and recover from mistakes. They work more like we do. The dif...

A discussion of the Deep Agents architecture with Sydney Runkle and @mkennedy.codes on @talkpython.fm

talkpython.fm/episodes/sho...

TUI interface to the Deep Agents based coding CLI, with text in greent, and a very pixelated parrot mascot. The agent says Hello, world!

Deep Agents is an open source project from LangChain to let anybody build their own model-agnostic agent harness (tools like Claude Code or Codex). Love that the mascot of their (proof-of-concept) CLI is a (presumably stochastic) parrot. 🦜

github.com/langchain-ai...

FERC Final Rule Reforms EQR Process and Reporting: Key Takeaways Nearly three years after proposing reforms, the Federal Energy Regulatory Commission (FERC) has largely adopted its proposed revisions to Electric Quarterly Report (EQR) requirements.  The proposed ru...

Lol, now that we've integrated EQR of course FERC has approved changing the format in which it is reported. Hopefully they'll do some more data validation? Maybe add some more real IDs for better record linkage? We'll see.

www.mcguirewoods.com/client-resou...

You too can check out the FERC EQR through the PUDL Data Viewer:

data.catalyst.coop/search?&q=fe...

A chart showing three different views of S3 usage over the last quarter, from the beginning of 2026 to the end of March. It shows daily unique IPs, GB of data downloaded, and normalized file downloads. The GB downloaded has a huge spike in the last week of the chart.

Wow. Last week a bunch of new people discovered our free, public FERC EQR Parquet dataset on S3. It details billions of US electricity transactions covering 2013Q3-2025Q4. They downloaded more than 2 TB of data.

#EnergySky #OpenData #Electricity

Hi #EnergySky! Tomorrow's the deadline to apply for our data lab this summer - if you've been putting it off, please stop doing that and fill out an application!

But this is more than a discussion about a beautiful, performant static site generator -- the "Zensical Spark" funding + user engagement model is really similar to what we want our PUDL Sustainers program to be. Great to see that someone is already making it work.

Zensical - a modern static site generator If you've built documentation in the Python ecosystem, chances are you've used Martin Donath's work. His Material for MKDocs powers docs for FastAPI, uv, AWS, OpenAI, and tens of thousands of other pr...

Really good discussion between @squidfunk.com & @mkennedy.codes on @talkpython.fm about the origins, goals, and status of @zensical.org Looking forward to trying it out and looking at trying to migrate our docs from Sphinx once it is more fully baked. #Python #FOSS

talkpython.fm/episodes/sho...

#EnergySky folks! There's a week left before the application deadline for our energy lab this August. Now is the time!

The Utility Business Model Is Built for a Different Era. Regulators Are Starting to Notice. Post 1 of 10: The Potential Structural Transformation of the U.S. Electric Utility Industry

Interesting series of posts on the utility political, regulatory, & business model landscape, how & why they're failing to adapt, and how changing economics of DERs interact with the system. From the former CEO of Octopus Energy.

#EnergySky

distributedgrid.substack.com/p/the-utilit...

CMU Energy Week The Wilton E. Scott Institute for Energy Innovation's flagship event, CMU Energy Week, brings energy and sustainability leaders, including scholars, investors, and entrepreneurs from across the nation...

Catalyst member Kathryn Mazaitis will be at CMU Energy Week this week to connect with students and other potential PUDL data users. Come say hello! @costasamaras.com #EnergySky

energy.cmu.edu/energy-week/...

For example, here's Xcel Energy Colorado's Cherokee plant generation by fuel telling the tale of its conversion from coal to gas. And then also total generation broken out by generator.

Right now it only runs 2006-2021 but if it's useful, we could look into what it would take to make new FOIA requests and get it updated annually.

We're developing some interactive, in-browser @marimo.io #DataViz notebooks in the PUDL Data Viewer. Definitely not ready for prime-time, but if you're up for some alpha testing, we'd love feedback. The first one lets you explore plant-level EIA data. #EnergySky

data.catalyst.coop/secret-examp...

Hey #EnergySky we have a new monthly PUDL release v2026.3.0.

We've *finally* gotten the last EIA-923 8 table (emissions controls) in. Also a *ton* of new Rural Utilities Service data from FOIA requests.

Release notes:

catalystcoop-pudl.readthedocs.io/en/nightly/r...

GitHub - tobi/qmd: mini cli search engine for your docs, knowledge bases, meeting notes, whatever. Tracking current sota approaches while being all local mini cli search engine for your docs, knowledge bases, meeting notes, whatever. Tracking current sota approaches while being all local - tobi/qmd

Another locally focused project that uses small models and semantic search for your own documents

github.com/tobi/qmd

Ask AI | Transform your CKAN data portal into an insight engine YouTube video by Link Digital

Video demo

www.youtube.com/watch?v=yZcy...

Ask AI: Link Digital’s AI layer for smarter CKAN portals Ask AI introduces semantic similarity search. Instead of matching words, it matches meaning. Read more about Ask AI by Link Digital.

Paper about the project

linkdigital.com.au/news/2026/03...

Ask AI - Link Digital Transforming your CKAN-based data portal into an insight engine. Ask AI. Get in touch with Link Digital.

Ask AI is aimed at privacy-first queries -- supporting local LLMs and local data, smaller models that are specialized to task

linkdigital.com.au/ask-ai/

Exploring Data Portals with AI: Queryless + PortalJS Demo YouTube video by Datopian

Video demo of Queryless

www.youtube.com/watch?v=UBT5...

Queryless — Talk to Your Data, Get Dashboards Fast Ask questions in chat and get trustworthy answers plus interactive dashboards in under a minute.

Queryless is another natural language query project, including over instant messaging apps like Telegram.

www.querylessai.com

Recorded DataJourney lightning talk (Sayantika couldn't make it -- this is an extremely global gathering) drive.google.com/file/d/1cQNu...

GitHub - DataJourneyHQ/DataJourney: Design first Open Source Data Management Toolkit Design first Open Source Data Management Toolkit. Contribute to DataJourneyHQ/DataJourney development by creating an account on GitHub.

Project repo here: github.com/DataJourneyH...

DataJourney is an open source platform for querying and visualizing data with natural language

datajourneyhq.com

Some other examples of interesting #OpenData + #LLM applications... #AI