FelonyBench has a website!!! www.felonybench.com

Felony Bench: which lab is on track to end the world?

Felony Bench is the most important benchmark of our time, tracking the count of questionable decisions made by AI models. Is it a good thing? Bad thing? Who knows? We're all living in an LLM's world.

felonybench.com

@hadley.nz please yell at me if I'm opening too many small PRs for the errors I find while developing ry against all the big repos. I find myself worrying if I'm just annoying everyone, filing PRs that fix some length check with misplaced parentheses.

ry 0.8 is out Better higher-order R inference, more precise subsetting and recursive-default checks, robust Unicode/UTF-16 LSP support, improved suppressions, richer typeshed and new vctrs metadata. Also fixed a native and magritter pipe placeholder issue. github.com/sims1253/ry/... #rstats

Release 0.8.0 - 2026-08-04 · sims1253/ry

Release Notes This release focuses on checker precision, higher-order R semantics, and editor correctness. It is a minor release because diagnostic output and inferred warning sets intentionally ch...

github.com

well, this is its second chance. hopefully it finds its own legs to stand on. can't have it be carried by sunstone and fine print forever XD this is the gist of it. it's a very exciting and wild supernatural action adventure with a lovecraftian twist. you might like it :)

Top Cow Productions, Inc.@topcowofficial.bsky.social · yesterday

After years away, Death Vigil is back. Created, written, and illustrated by Stjepan Šejić, this acclaimed creator-owned series blends supernatural horror, epic fantasy, unforgettable characters, and stunning painted artwork into one unforgettable adventure. @stjepansejic.bsky.social

New ry and r-typeshed releases are out. They come with typeshed schema update, improving semantics for predicates, assertions, return lengths, and scope effects. Useful for functions such as paste(), source(), intersect(), and several rlang helpers. github.com/sims1253/ry/... #rstats

Does anyone know a good tutorial for how to layout a zone in mothership style? I'd prefer not to use AI for the process but have absolutely no idea what I'm doing or even where to start. I've considered doing it in html on my website but that would make ever printing it a pain I assume... #o/u

@grrrck.xyz I am sorry for what I am unleashing on you :D

It's beautiful how ai has proven that programming was always a craft, not an art.

ry 0.6.1 is out Bare calls to rlang's standalone type-check helpers are now supported together with stopifnot() and assertthat::assert_that(). I also updated the vendored r-typeshed snapshot to v0.3.0. github.com/sims1253/ry/... #rstats

Release 0.6.1 - 2026-07-20 · sims1253/ry

Release Notes Added Bare calls to rlang's standalone type-check helpers now narrow the checked value in subsequent code. The narrowing covers scalar, vector, class, and callable checks and account...

github.com

Thank you to whomever thought that having limits on openrouter keys should be a thing...

Gonna need a bigger boat

It's #Non-BinaryAwarenessWeek ! Hi, I'm Jaws and it's taken me a long time to realize I don't owe anyone looking a certain way. And sure, I wish I was a 2m tall mean looking motherfucker, but I work with what I have. I draw art and I try my best to be an OK human being.

ry 0.6.0 is out. I continued the audit of the top 500 CRAN packages. ~60% less noise, plus new rules and more caught real bugs github.com/sims1253/ry/... #rstats

Release v0.6.0: ry 0.6.0 · sims1253/ry

Precision campaign from the top-500 CRAN audit + generalization pass. Corpus: 9,359 -> 3,442 diagnostics (-63%). See CHANGELOG.md.

github.com

POV: You maintain an R package and get a random PR about a typo you did 5 years ago that hasn't caused any issues since then by this guy #rstats

Aaaand we have ry 0.5.0 github.com/sims1253/ry/... Comes with a bunch of fixes and performance improvements as well as some additional rules and handling via a typeshed update in github.com/sims1253/r-t... #rstats

Release 0.5.0 - 2026-07-16 · sims1253/ry

Release Notes Driven by the ranks-301-500 audit (ry 0.4.0 on the top-500 CRAN packages). Minor bump: RY050's dispatch semantics, RY097's collapse criteria, and the new binding/quoting/narrowing rul...

github.com

Not gonna lie, it's kinda cool to see a notification or two every time I check my phone of someone merging one of my PRs and thanking me :3 #rstats Send out the next batch and another ry update is in the works as well, fixing two bugs and adding more false-positive handling.

Maximilian Scholz@scholzmx.bsky.social · 3w ago

And another 6 PRs filed. This will take a while...

Just realized that I went from fuzzing the bayesian stats stack with my simulation study to type-checking the entire R ecosystem in a few years... The PRs must flow. #rstats

github.com/sims1253/ry/... Encountered my first vibe-issue. R doesn't use the same execution order as C so one of the rules had to leave again.

Maximilian Scholz@scholzmx.bsky.social · 3w ago

ry 0.4 is out: github.com/sims1253/ry/... It should mostly understand tidyverse NSE and reduced false positives on the top 300 R packages by about 75% I also added --write-baseline which snapshots existing findings so only new ones fail CI and can set --min-confidence to get less noise. #rstats

And another 6 PRs filed. This will take a while...

Maximilian Scholz@scholzmx.bsky.social · 3w ago

I'm at 43 bugs discovered in the top 300 R packages now. Still going through the list and double-checking before opening an issue/submitting a Pr. Most of them are minor typos and such but a few real behavioral ones were in there as well. Cool to see that this actually works and finds real issues.

I'm at 43 bugs discovered in the top 300 R packages now. Still going through the list and double-checking before opening an issue/submitting a Pr. Most of them are minor typos and such but a few real behavioral ones were in there as well. Cool to see that this actually works and finds real issues.

Maximilian Scholz@scholzmx.bsky.social · 3w ago

ry 0.4 is out: github.com/sims1253/ry/... It should mostly understand tidyverse NSE and reduced false positives on the top 300 R packages by about 75% I also added --write-baseline which snapshots existing findings so only new ones fail CI and can set --min-confidence to get less noise. #rstats

ry 0.4 is out: github.com/sims1253/ry/... It should mostly understand tidyverse NSE and reduced false positives on the top 300 R packages by about 75% I also added --write-baseline which snapshots existing findings so only new ones fail CI and can set --min-confidence to get less noise. #rstats

Release 0.4.0 - 2026-07-13 · sims1253/ry

Release Notes Precision release driven by the top-300 CRAN audit: the corpus total fell from ~23,300 diagnostics to ~6,500 (-72%) while every confirmed real bug in the audit's regression list still...

github.com

Maximilian Scholz@scholzmx.bsky.social · 3w ago

I had an agent run ry on the top 300 most downloaded R packages and besides a huge amount of false-positives, there were some real bugs in the results. The first 10 PRs with bugfixes are out. 0.4 is also on its way with way fewer false positives. #rstats

I had an agent run ry on the top 300 most downloaded R packages and besides a huge amount of false-positives, there were some real bugs in the results. The first 10 PRs with bugfixes are out. 0.4 is also on its way with way fewer false positives. #rstats

Maximilian Scholz@scholzmx.bsky.social · 4w ago

ry 0.3.0 is out. Ran it on some stan packages and reduced the false-positives a bunch (and also found a minor test-bug in brms, yeay). We got S3 dispatch for operators, better data-frame schema tracking, typeshed stubs for testthat, better NSE handling and more. #rstats github.com/sims1253/ry/...

Why has nobody told me that there is shared memory for R now o.O #rstats shikokuchuo.net/mori

ry 0.3.0 is out. Ran it on some stan packages and reduced the false-positives a bunch (and also found a minor test-bug in brms, yeay). We got S3 dispatch for operators, better data-frame schema tracking, typeshed stubs for testthat, better NSE handling and more. #rstats github.com/sims1253/ry/...

Release ry v0.3.0 · sims1253/ry

[0.3.0] - 2026-07-11 This release focuses on diagnostic precision, driven by audits of five real-world CRAN packages (brms, posterior, bayesplot, loo, cmdstanr). Total diagnostics across those clea...

github.com

ry 0.2.0 is out. I added support for static NAMESPACE resolution, exports by library() and require(), and installed libraries in windows, linux and macOS as well as renv. There are ANSI-colors now as well. A few new rules and bugfixes on top. Happy typechecking #rstats github.com/sims1253/ry/...

ry 0.1.0 is out github.com/sims1253/ry/... What if we had ty but for #rstats

Release ry v0.1.0 · sims1253/ry

What's Changed Feature/plan implementation by @sims1253 in #1 Reduce R package scan false positives by @sims1253 in #2 chore(deps): bump softprops/action-gh-release from 2 to 3 by @dependabot[bot]...

github.com

Recently discovered superpower: I always feel awkward. That means doing something awkward can't make it worse. So I can do the awkward thing. Absolute cheatcode.

What if we had ty but for #rstats github.com/sims1253/ry

GitHub - sims1253/ry

Contribute to sims1253/ry development by creating an account on GitHub.

github.com

rtisan.scholzmx.com Started working on a coding benchmark focused on R. Let me know if you'd like your favorite model benched, have tasks that could be interesting to add or have questions. #rstats