fucory

Proudly introducing JJHub A new code platform for agentic engineering teams 🧵 1/n

Breaking! Anthropic acquires pip install

@krzKaczor @continuedev I got a workflow to validate a bug is real and then comment on the issue

@FUCORY AI powered CI checks are really intriguing. Recently I've started exploring @continuedev. For now I use them mostly to enforce again some things from AGENTS.md. Do you have some inspiring examples?

The power of Smithers is gonna really be unlocked when people start combining it with jjhub JJHub is like if we took github actions, replaced yaml with typescript, and made the workflows super fast to run locally and in the cloud. In this world you want your workflows to be composable reusable and easy to improve over time. You start with a simple set of high impact workflows - ReproduceBug - CIChecks - VerifyDeploymentSuccess - ReviewPr You run these checks developing with clankers both locally and in cloud Engineers invest time into improving the workflows over time. They factor out reusable pieces to build more sophisticated workflows. Just one step at a time. Slow careful evolution 12 months later that engineering team will have a secure, robust AI automated set of workflows improving the velocity and quality of the codebase unlike any engineering teams before

@EricForgy @matthuang Too expensive to get me to stop building jjhub.tech

@matthuang @FUCORY FUCORY is the 🐐 Aquihire is the right move.

I feel 100% confident Tempo will ship state channel solution or something of similar flavor Why? Because if Tempo doesn't ship a state channel solution I don't trust they actually know what they are doing. But I would bet they know what they are doing

@sameandnot @agenticklabs Yea you can also use the CLI agents like Pi Amp ClaudeCode Opencode or Codex Your API got some things I should have been doing better so if you notice anything where you think your approach was much better open issue and I'll definitely make any changes you suggest

@FUCORY @agenticklabs This looks really cool. I’m not very familiar with JJ but from what I understand of it this approach makes a lot of sense. Agents are run with ai-sdk under the hood, yeah?

It sounds so gimmicky AI Orchestration framework via React But hear me out, Smithers is secretly the best-in-class agentic orchestration framework In this 🧵 I answer two questions 1. Why you even need an orchestration framework? 2. Why that orchestration framework should be smithers 🧵 1/18

@samuellhuber @MiniMax_AI @ryancarson 👀

@MiniMax_AI @ryancarson in case you're code factory doesn't do this, it's likely a helpful overview for you cc @FUCORY re agents reading transcripts + metrics from smithers flows, updating the workflow / agent (e.g. pi extension writing on the go) -> it hot reloads and they watch for eval.

Introducing MiniMax-M2.7, our first model which deeply participated in its own evolution, with an 88% win-rate vs M2.5 - Production-Ready SWE: With SOTA performance in SWE-Pro (56.22%) and Terminal Bench 2 (57.0%), M2.7 reduced intervention-to-recovery time for online incidents to 3-min on certain occasions. - Advanced Agentic Abilities: Trained for Agent Teams and tool search tool, with 97% skill adherence across 40+ complex skills. M2.7 is on par with Sonnet 4.6 in OpenClaw. - Professional Workspace: SOTA in professional knowledge, supports multi-turn, high-fidelity Office file editing. MiniMax Agent: agent.minimax.io API: platform.minimax.io Token Plan: platform.minimax.io/subscribe/toke…

@__Raiders @matthuang *acquire

@matthuang @FUCORY hire him matt

I imagined myself working there. I'd think about the trust assumptions of stablecoins. I'm generally a product first person. And state channels produces the best product. I would have felt very strongly about this to the point where if I couldn't convince leadership that this is the correct path I would lose faith in leadership and leave So that combined with knowing leadership does in fact know what they are doing made it clear to me it's inevitable

@FUCORY How did you know?

@SCBuergel @samuellhuber 🫡

@FUCORY @samuellhuber Thank you for keeping the spirit alive 🙏

A billion Wei idea for all you fancy VCs here: An Ethereum node that just works. Not stateless. No GigAgAss. No L2. No zk. But an unbelievable killer feature that nobody tried yet: You can query state and logs and the node (a) always responds with sub-minute latency and (b) actually responds correct data. I'm happy to invest the first 10k Wei myself.

I was an external contributor to that team and one of their biggest users. Because of the team getting laid off that is what forced me to build Guillotine before guillotine I was using ethereumjs as my evm implementation That plus performance was a 2nd reason I needed to build it

@FUCORY @samuellhuber Interesting, thanks for that. Suggestion: add this to your bio - it's hard to find here. IIRC there was some drama around the Ethereum JS team getting laid off from EF last year, is that the revival of those efforts or entirely separate?

@sameandnot @agenticklabs Very cool! I'm building jjhub.tech to try to get this type of orchestration running in easy to manage containers in an alternative github. If we got similar interests might be an opportunity to collaborate

@FUCORY @agenticklabs I did… Smithers likewise looks very cool! And I love the name. :) When I said “should have” what I really meant was “wish you had”! It’s exactly the kind of higher order system I built agentick to be able to support. I am using it to build github.com/agenticklabs/t… WIP

@sameandnot @agenticklabs Looks cool! I think smithers is a bit of a higher level of abstraction than this. This looks useful for managing context windows at a low granular precise level and likely would work with smithers with minimal effort Did you build it?

@FUCORY Should have just built it on top of @agenticklabs github.com/agenticklabs/a…

tevm.app is an abandoned project. A block explorer I built in 2 days using voltaire The Tevm universe is vast We got the original tevm: github.com/evmts/tevm-mon… We got voltaire: github.com/evmts/voltaire We got guillotine and guillotine-mini: github.com/evmts/guilloti… We got zevm (unreleased but likely next week) which includes light client consensus and lightweight way to sync and index There is also friends of tevm @trueblocks that has a powerful indexer I used in my prototypes of a local first fast syncing ethereum node

@FUCORY @samuellhuber I'm confused what I'm looking at tevm.app Docs links all send me to some generic Base docs. How do I run this locally on my laptop?

@shazow Actually nvm only the people who bet on openclaw are rich

@shazow My longing of electrobun 2 years ago and Effect.ts 4 years ago would have made me rich

Understandable! I spent a lot of time playing with a builder pattern, Effect.ts, a python decorator pattern, and more and tried really hard to find any pattern better than JSX. They were all not a peak as JSX though The exception is TOON works great for simple scripts so we support that

@FUCORY sorry i just get PTSD looking at react syntax

my eyes have been cursed

The key is to just constantly look out for repetitive things you can abstract. Start with things smaller than smithers. Maybe set up a postcommit hook to have an llm review the code. Maybe use bun to just create a super simple script or make some custom pi plugins. Abstract prompts you repeat a lot into reusable skills. As you climb that ladder of automation eventually the problems smithers solves will appear and why you should use it becomes obvious

@FUCORY that said, i guess the next level would be - defining bigger (and/or more) tasks so i have to manually run less often the loop, and handling to agents - having a proper tasks system but still no idea how to make smithers fit here

@kartojal Yes I've built this workflow using linear in past

@FUCORY can i use smithers to watch Jira tasks and proceed to plan the task, them work (code) on it in autopilot, test it, and review the output too?

@0xDrewF @powdr_labs @openvm_org Whoah that sounds cool. I would have to look into it but that would be awesome feature

@powdr_labs @openvm_org Does this mean proofs of Voltaire are straightforward now? @FUCORY For example I could prove there are X logs that match a filter over some blockrange?

Announcing powdr-wasm! powdr-wasm is an optimized zkVM for WASM, built on top of @openvm_org and the novel 𝑐𝑟𝑢𝑠ℎ ISA. Early benchmarks already show 1.5x fewer trace cells & faster proof times compared to RISC-V (OpenVM). It also supports Go guests via WASI! 👇