dr phosphorus

@mehmetozsoyart It’s beautiful man.

Finally finished the hero animation for Mendly. Built around an isometric AI driven care flow system that visualizes automation from referral to final claim.

@doodlestein what are you recommending for best repo wide clean up. Getting rid of the Claude clutter? Or do you recommend softer optimization at each function?

@JameyBaskow Tippet should not miss 1 more game for that. Standard surgery.

If Owen Tippett indeed has a sports hernia. Honestly, I do not expect much of him until February of next year. The Flyers have had bad luck with these surgeries (Drysdale, Girouux, Ghost, Hates), most recently Drysdale. Drysdale wasn't 100% until February of last year

@alxui_ux by who? trying to find more screens of this. looks beautiful

Soft gradients + mindful UX = perfect mobile aesthetic ✨ By @woinlesss

@AravSrinivas Is this available as an API?

Perplexity and Computer now allow you to run Deep and Wide Research on sources trusted by doctors and medical professionals like the New England Journal of Medicine, the British Medical Journal, the American Diabetes Association, and so on.

@DR3Wheels Honestly he's overplayed the 4th line. #41 was atrocious during the game the other night and has no right being out there when the team is down.

Thought about the Michkov scratch a little and here's where I'm at: I do want to get Bump in, and for that to happen someone needs to come out. It's not going to be a center. It's not going to be Martone, Tippett, or Konecny. It's not going to be anyone on the 4th line. That leaves Foerster, Barkey, and Michkov. All have had their struggles this series. Foerster is still sound defensively even if his offense has been meh at best. I don't think you can take him out with the game plan they've been running all series. Putting him with Cates again will help him find his offensive game too since he doesn't have to be the primary defensive guy on his line. That leaves Barkey and Michkov. They both have been underwhelming and looked a little small out there but not to an egregious extent. The difference between them this series is two things: 1) Barkey has scored a goal and Michkov has not 2) Michkov has taken a LOT of stupid penalties, some have ended up giving a goal back to the Penguins (Game 3 roughing leading to a Karlsson PP goal as an example) I think the Michkov scratch also isn't because it's a lack of trying or anything physical like that, I think he just needs a mental reset. You can see him getting frustrated out there, and even see boiling over to dumb penalties. I really don't think a one game scratch to get his head right is a big deal. Michkov will be fine, the Flyers will be fine, let's go win a hockey game tonight.

@doodlestein Not to mention keeping it up to date.

@doodlestein I just did this doc skill and it’s wildly impressive. Production quality documentation better then most saas offerings. So good

I wrote an article about how to use my new will writing and estate planning AI agent skill that breaks it down for non-technical people. It’s very easy to use now and just requires a few clicks to get started with the new desktop Claude and Codex apps: jeffreyemanuel.com/writing/wills-…

@doodlestein I have a 50 person team in ops and 5 are dedicated to docs. Engineering has a 300 person team and cannot deliver anything like what your skill just produced. I promise you this.

@hxxwhite How are your opening it up in codex? Just a url?

Mobile dev in 2026 looks nothing like mobile dev in 2024. Editing an iOS app with Codex, live, in a real iOS simulator running in the cloud on Revyl, with Expo hot reload OTA. Every change shows up instantly. No Xcode. No provisioning. No cables. Whole app took ~20 min start to finish. Not a single line of code written.

@doodlestein Trying it now

I finally got around to making a skill a lot of people have been asking me for: jeffreys-skills.md/skills/simplif… It basically helps to "de-slopify" and refactor code that's been written by agents, looking for ways to simplify and reduce the amount of code without changing the behavior. The difference between this and other skills or prompts in the same spirit is the lengths this one goes to in order to prevent the process from going off the rails and introducing bugs or security problems. It's a whole elaborate system spanning 98 files and one full megabyte of reference files, scripts, and subagents (see pic). You can run it over and over again and it will autonomously identify good opportunities for accretive simplification and do everything needed to implement the changes and prove that they didn't change the outputs. GPT-5.5 can explain better than I can how it does all that and what makes it so compelling and useful: --- The strongest thing about this skill is that it treats refactoring as a proof obligation. A normal “clean this up” prompt invites the model to follow taste. It sees repetition, long files, wrapper functions, stale types, try/catch clutter, _v2 files, and it starts cutting. Sometimes that works. Sometimes it silently changes error semantics, loses a side effect, removes a lifecycle hook, or deletes a file that looked unused but was actually the intended implementation path. This skill changes the frame. A simplification claim becomes: “this smaller program is observably equivalent to the larger one.” Then it makes the agent prove that claim. It starts with a baseline: tests, golden outputs, LOC, warnings, complexity. It maps duplication instead of eyeballing it. It classifies clones, because exact copy-paste, parametric duplication, semantic similarity, and accidental rhymes are completely different things. It scores each candidate by expected LOC saved, confidence, and risk. Low-score candidates get rejected and logged, which is important because future agents otherwise rediscover the same bad idea forever. The isomorphism card is the key move. Before editing, the agent has to answer boring but lethal questions: same ordering, same errors, same logs, same metrics, same side effects, same async cancellation behavior, same React hook identity, same serialization, same resource lifecycle. Those rows catch the kind of bugs that compilers and ordinary tests miss. Then the edit discipline is deliberately narrow: one lever per commit, no rewrites, no sed, no drive-by fixes, no deletion without explicit permission. Afterward, it verifies behavior again and records the result in a ledger. If the refactor did not actually preserve behavior, it does not get to call itself a refactor. What I like about it is that it matches the real failure modes of agent-written code. AI code tends to accumulate plausible junk: defensive branches for impossible inputs, duplicated wrappers, too many optional parameters, orphaned “improved” files, shallow happy-path tests, stale types, and comments that are really leftover task plans. The skill has a whole pathology catalog for those patterns, plus scripts and subagent roles to find them systematically. So the compelling part is not “make the code prettier.” The compelling part is leverage with brakes. You can send very strong models into messy codebases and ask them to reduce complexity aggressively, while forcing them to preserve the contract that matters: observable behavior. That is the difference between a refactor you hope is safe and a refactor you can audit.

@ClaudeDevs Need to refund people.

Over the past month, some of you reported Claude Code's quality had slipped. We investigated, and published a post-mortem on the three issues we found. All are fixed in v2.1.116+ and we’ve reset usage limits for all subscribers.

@TUSHARPAND0848 Left

Be honest, left or right?

@ecommerceshares I had the same thing. 3 different max subscriptions. Something is wrong. They need to do massive refunds.

Claude Pro sub. Ran a Sonnet 4.6 prompt twice. 2 times. Not even on Opus. Immediately hit my session limit and 43% of my weekly limit. What??!?!

@ReadySetBrian @bcherny I canceled my 3 max subscriptions as well. Something changed the last week as well. My usage burnt out within a few days, all 3 at the same time. Brutal with no explaination.

Canceled Claude max today, @bcherny whatever happened in the last 1-2 months is a significant regression. The model feels like someone from OpenAI started working on trust and safety there. Opus thinking is significantly worse. Every statement is “here’s where I’d push back on that” and then proceeds to rattle off the most inane list of confused counter arguments. It was perfect 3-4 months ago!!!

@TheGeorgePu They did something with max too. I hit my weekly limit on all 3 subscriptions today. 2 days in advance. First time that’s ever happened. I’m done.

Anthropic just pulled Claude Code from the Pro plan. Pro users wanting it need Max now. $100/month minimum. 5x jump. I'm on Max 20x so I'm fine. Flagging for anyone on Pro who's about to find out. No announcement. Just a pricing page edit.

@rohanvarma After having 3 max subscriptions since October and then all hitting their limit suddenly today all at once, I’m ready to switch. Any tips on the switch?

For clarity, we are running a small test for ~100% of Codex users where we: - make our best models available to all users - make Codex available to all plans, free and paid Claude Code users aren't affected :)

I’ve had 3 separate Claude max subscriptions running monthly since October. ($600+ a month). For the first time since then, they all hit their weekly limit 2 days in advance of reset and at the same time. This was on a week with light usage, on Opus 4.6 (I haven’t upgraded yet). Anthropic is pulling rugs and the lack of transparency is egregious.

@basit_designs @joshbuildings this isnt even claude design interface

Exploring a little effort with claude and this what it generates. No design tool used, no code harm, just few lines of prompt and here the result