Yes, that’s part of the goal. It does not force the LLM to pick the LSP tool so it may still grep, but it gives it the option to use it. It will also automatically give it feedback as code changes, which should help it get the task correct without needing as many follow-up prompts
Here's a quick guide to setting up an R LSP for Claude Code in #rstats: christophertkenny.com/posts/2026-0...
Note that it uses the `languageserver` package at the moment, since I don't believe ark is exposed for a standalone mode yet.
I see the appeal for a lot of social scientists. There are many folks in the field who are consumers of data analysis/statistics because they see it as necessary to publish research.
Many use RStudio and have no interest or experience using other IDEs or the CLI. This serves that group well.
News for #RStats folks who use RStudio and are interested in AI, via @posit.co: "AI in RStudio is here. We’ve embedded a specialized agent, Posit Assistant, directly into RStudio, transforming the IDE into a powerful, context-aware collaborator for data analysis."
It's a $20/mo subscription.
Over the weekend, I started collecting/organizing a list of cool #rstats skills for LLMs at github.com/christopherk.... Please open an issue or PR in any awesome additions you see!
Includes a section for helpful posts like @ivelasq3.bsky.social's⬇️
One weird pattern from playing with Claude Code and GitHub Copilot CLI:
Claude Code loves to use bash and python, even when working on R code (and then gets confused since there is no 🐍installed)
Copilot, with the same models, seems to naturally know to use R and powershell
I moved everything CRAN-related to one proton mail (to avoid changing academic emails) and it’s incredibly nice to have it all in one place without any other noise. Much easier to keep track of deadlines…
Getting ready for the 2026 mid-terms, every PlanScore analysis page moving forward will now feature a new scenario adjustment tool to make it easy to change, view, and share potential outcomes that may differ from 2024
medium.com/planscore/pl...
Try with this NC map:
planscore.org/plan.html?20...
Thanks Rohan! Hope you’re doing well
Read the preprint here: arxiv.org/pdf/2407.11336
Redistricting Reforms Reduce Gerrymandering by Constraining Partisan Actors is now forthcoming in the APSR!
We use a game theoretic treatment and continuous DiD(iD) to show when redistricting reforms work
with @corymccartan.com, @simko.bsky.social, Emma Ebowe, Michael Zhao, and Kosuke Imai
My 2025 #rstats wrapped in honor of CRAN reopening for 2026: christophertkenny.com/posts/2026-0...
Fun fact: I made 43 CRAN uploads in 2025 and 41 of them were successful!
Throwback to my favorite #rstats Ripley crashout where he removed a package because a maintainer called him Brian
Some people are fundamentally unkind. It's a shame. The broader R community is incredible and the other CRAN volunteers especially so.
Nothing like starting the day responding to a Brian Ripley hissy fit...
He removes an #rstats package before the scheduled date, gets a bounceback from a *downstream* package, and archives all of the packages...
CRAN volunteers time is the most valuable resource, yet Ripley just loves wasting it
Oh awesome, thank you both. I had looked for this a few weeks ago and didn’t see an option
Glad we have LLMs in everything so that GitHub tries to create commit messages like this... Just makes more work by having to delete the message to write in a correct one instead of just writing the correct one in the first place
Is there an #rstats best practices for including a shiny app in an R package? I see mastering-shiny.org/scaling-pack... by @hadley.nz and github.com/daattali/ddpcr by Dean Attali (couldn't find his handle!) - just curious on understanding best practices, bunding data, app, and the www folder etc.
@corymccartan.com and I have been playing with @paleolimbot.bsky.social's #rstats `libgeos` package. It's pretty awesome.
Handles the hard parts so that you can connect Rcpp to GEOS with a few lines of code. Got some existing, optimized `geos` code to ~3x faster!
paleolimbot.github.io/libgeos/
Oh awesome, thanks for sharing. Kieran Healy is always doing awesome work. Not shocked that he did this first!
A dot density map of NYC where each point represents 50 voters. Lines show NYC neighborhoods. Density is highest in Manhattan with a lot of dots for Mamdani. Areas of Staten Island and Brooklyn are dense and yellow for Cuomo.
NYC 2025 Mayoral results as a dot density plot using the `dots` #rstats package (cran.r-project.org/web/packages...)
Do you teach #rstats? Do your students complain about how lame and old-fashioned dplyr is? Don't worry: I have the solution for you: github.com/hadley/genzp....
genzplyr is dplyr, but bussin fr fr no cap.
Support for Prop 50 in California is highly correlated with support for Harris in 2024 and even runs slightly ahead of her
A sad day for democracy to have the future of competitive elections literally on the ballot and yet be so utterly defeated.
Stands at 63.9% voting yes.
Thanks for sharing!
There are a lot of great posts out there that aren't very highly ranked.
Don't rely on bluesky to find you great content; you can find it on your own! Here's how:
#Rstats via @northeasternu.bsky.social's Storybench
www.storybench.org/how-to-analy...
Yeah and once you start adding known precinct covariate, the possibilities really explode. We ended up with several internal (quarto) dashboards to handle plotting everything on the fly
NYC map shaded by vote margin!
We collected very granular precinct data and aggregated it to neighborhoods people actually recognize
Bushwick, Flushing, Upper West Side, etc.
datawrapper.dwcdn.net/Wrd72/
H/T @mitelectionlab.bsky.social @chriskenny.bsky.social @corymccartan.com @simko.bsky.social
For the on-air presenting and digital media, we also built a live neighborhood map: www.cbsnews.com/news/nyc-may... (now filled by margin!)
Here at the CBS News data desk with @chriskenny.bsky.social and @simko.bsky.social! Looking at the VA numbers
Finishing up prep for election night at CBS News' Data Desk!
Sunday in NYC was the highest single day early votes since at least 2019 and maybe ever:
