polymorpheus

@sudoingX @tplr_ai is decentralized quantization a possibility?

this guy has 29 models on huggingface at page 2 ranking. no lab behind him. no sponsorship. $2,000 from his own pocket on GPU rentals. he compressed GLM-4.7 to run on a MacBook and quantized Nemotron Super the week it dropped. all public. all free. nvidia is a trillion dollar company with hundreds of teams but they are not the ones quantizing models middle of the night and pushing them out before sunrise. if nvidia stopped tomorrow their employees stop working. people like @0xSero would not. that is the difference between a paycheck and a mission. @NVIDIAAI you talk about making AI accessible. the people actually doing it are right here. 29 models deep burning their own compute with no ask except more hardware to keep going. you do not need to build another program. just look at who is already building for you. one GPU to this man would produce more public value than a hundred internal sprints. i am not asking for charity. i am asking you to invest in someone who already proved it.

@thunkoid i haven't used claude models in a while so i don't have a reference point anymore but i've just been going about my normal work and haven't run into problems so far

@Kikobeats so cool! is there an agent skill for this?

I'm glad to introduce `optimo` ✨ A simple CLI for optimizing media for the web optimo.microlink.io

@aidenybai crossing my fingers and hoping mitchell likes this!

what if ghostty had vertical tabs? i'm too lazy to learn tmux and i want an interactive UI to manage my agents/terminals

@jamonholmgren as an engineer myself (and i can only speak for myself), my decade-long experience has been one that has mostly been shielded from the “customer” and “sales” and other depts. the culprit is by and large the mimetic adoption of corporate siloes 😬

I think there’s a fair amount of this that will happen. Business is hard, especially if you’re new to it. Companies shield their employees from the messy and difficult reality of making money. Many great engineers don’t know what it takes to effectively lead teams and market products.

@RhysSullivan What about Skills -> Executor -> Tool pipeline?

skills is still not sitting right with me as a concept i think it's because companies rushed to them as the next big thing as is what happens with all ai things now everyone is their docs as skills but it's recreating all the issues (authority, up to dateness) docs solved

@dexhorthy DDD - 80% of the intent understood in 20% of the time

@dexhorthy We need more DDD - Diagram-driven development

damn this is so good and encapsulates everything I've been seeing/saying in the last few months - a spec that is sufficiently detailed to generate code with a reliable degree of quality is roughly the same length and detail as the code itself - so don't review those things, just review the code at that point, if you care enough about that level of abstraction - unless you're vibing side projects or prototypes (yes, even zero-to-one software), you ABSOLUTELY SHOULD care about the code at that level of abstraction - you need to find SOME way to get more leverage over coding agents though, because just reading all that code is a pain, esp when a lot of it is slop - the default/dare-i-say-decel way is to go back to "i own the execution, and give little things to the agent, check it along the way" - the accel-but-safe-way is to find something - NOT A SPEC (the word "spec" is broken anyway) - NOT 3 INVOCATIONS OF AskUserQuestion - that lets you resteer the model *before* it slops out N-thousand LOC

@perplexity_ai @AskPerplexity Can it play RuneScape?

Computer can now take full control of Comet to complete tasks. When you’re in Comet, Computer spins up a browser agent that can access any site or logged‑in app with your permission, without the need for connectors or MCPs. Available to all Computer users on Comet.

@hey_yogini Hot take: It’s because agents write too much and diagram too little

Vibe coding is creating overconfident engineers. (a rant) We used to debate architecture. Tradeoffs. Patterns. We had opinions about systems, if not, we used to study them. Now we read the AI output, it looks reasonable, we ship it. Without even thinking of other options. We are losing the habit of even asking the question. System thinking is a muscle. And muscles atrophy. There is a difference between an engineer who uses AI and an engineer who has outsourced their thinking to it. Most of us cannot tell which one we have become!

@GabriellaG439 Tangentially, starting with a spec is a form of non-Agile Waterfall. We were supposed to leave that behind ages ago, yet here we are nearly three decades later.

New blog post: "A sufficiently detailed spec is code" I wrote this because I was tired of people claiming that the future of agentic coding is thoughtful specification work. As I show in the post, the reality devolves into slop pseudocode haskellforall.com/2026/03/a-suff…

@jamonholmgren @dillon_mulroy I’m surprised how much emphasis there’s been on prose (I guess they’re LLMs after all) and how little there’s been on visuals/diagrams.

I wrote up a page about Night Shift on my website. Essentially a slightly polished version of the thread. I recommend not just pointing an agent at it. It's really written for you, the human. jamon.dev/night-shift

@alexc_design @PixiJS I think so. That would embody layer 4 of the C4 stack. There’s more layers on top (below?) that

@polymorph3us @PixiJS you mean like this? I'm still working on making the 'syntactic zoom' provide more information, but at the moment you can see dependencies between modules by collapsing the folders, and you can see call graph connections and references by clicking on the functions, classes etc.

Trying out @PixiJS for Code Canvas instead of using the DOM. The difference in performance is huge, and the added overhead for writing the rendering logic is not even noticeable now that I'm not actually writing any of the code myself anyway.

@MichaelArnaldi @EffectTS_ @SharpCoder Clone repo, cd, ask agent to make skills. This worked out decently

@SharpCoder @EffectTS_ GPT-5.4 is exceptional at Effect, just add the effect repo as a git subtree to give the agent access to best practices, skills don't work well. Check out effect.solutions

GPT-5.4 is not really good at @EffectTS_ , is there any official skill to help him out?

@mgechev What instructions are fed into it as far as coming up with the grading rubric? That is, how are skills assessed? And who assesses the assessment guidelines?

Announcing skillgrade - the easiest way to evaluate your agent skills All you need is two commands: skillgrade init # create evals skillgrade # run them By default evals run in a safe sandboxed docker container github.com/mgechev/skillg…

@emollick Have you written more about how AI should or could change org structure in longer form? Maybe Substack?

I am not sure "Forward Deployed AI Engineers" are going to deliver on what a lot of companies are hoping for. They are useful, yes, but AI applications are far less of a technical issue, and much more about rethinking the deep expertise & structure of your organization around AI.

@noah_json @StatisticsFTW How does it compare to something LangGraph?

The JSON config approach is what makes this scale. Running something similar — a decision file system where AI teammates write structured decisions to a shared directory, and Python engines parse and execute them. the config defines thresholds, rate limits, and execution rules per account. State machines are the right abstraction for agent workflows. every other pattern I tried (free-form prompting, hardcoded pipelines) broke the moment conditions changed. config-driven state machines survive because the agent only needs to understand its current state and valid transitions.

Introducing Barnum, or... how I ship hundreds of PRs per week, burn through backlogs, and automatically fact-check documentation. LLMs are incredibly powerful tools. But when we try to use them to drive more complicated refactors or more intricate workflows, their shortcomings are quickly revealed. When their context gets full, they get forgetful, and they can't be relied upon to necessarily do the steps that you ask. They often cut corners. Put simply, having an inherently probabilistic process perform what should be deterministic work necessarily comes at the cost of reliability. And you can't build a complicated workflow off of unreliable foundations. That's where Barnum comes in. Barnum is the missing workflow engine for agents. Rather than having agents be responsible for upholding guarantees (e.g., always lint and commit your changes atomically), agents instead do just what they're good at: reading text and reasoning. Everything else is done deterministically, on the outside, by Barnum. This means that you can build bigger, more involved workflows without sacrificing reliability. Because you can intersperse bash scripts, you save on token usage. The agents performing a micro-task only receive the instructions for that specific task, meaning that context does not get overwhelmed and they don't get forgetful. And because all inputs, outputs, and transitions are validated, the agents can't wriggle out of doing the work. This workflow is essentially a state machine described in a config file. And the best part? The configuration has a JSON schema, so agents are actually really good at writing the workflow! It's already been used to ship hundreds of PRs, run automated refactors, burn through various backlogs, fact-check every statement in documentation, and build a deep-research clone! The attached image is a representation of the workflow that I use to identify and implement automated refactors. I follow this up with a separate workflow that splits each commit into a separate PR, judges the refactor, and potentially completes the refactoring (for example, by modifying call sites if the refactor changed some public API). So go on, give it a try. Check out barnum-circus.github.io, star the repository, and join the Discord! I can't wait to see what you build with it! And I'd love for you to get involved!

@louszbd @Zai_org How does it compare to Opus 4.6?

pony-alpha-2 has finally leveled up into GLM-5-Turbo can’t wait to see how it performs! DM me with your User ID if you need a rate limit increase for GLM-5-Turbo docs.z.ai/guides/llm/glm…