Jeremy Mack

@nutlope What about links in posts?

Prediction: AI-led codebases (like Codex and Claude Code) will become a named strategy employed by struggling SaaS companies

sometimes i get bored and go to a random tech company's website to see how they're pretending to be all about ai now

@ctatedev okay now you have my attention. this is very annoying in electron

Introducing Layered WebViews For website builders, browsers, others with preview surfaces in zero-native → Render previews in real WebViews, not iframes → No iframe cookie/auth/storage pain → Isolated by default → Bridge access only when trusted Web UI, stacked natively.

@richiemcilroy good intel. was considering an upgrade from the same chip, but in a mac studio

a couple of months ago, I upgraded from a M1 Pro 32gb RAM, to a M4 Max 128gb RAM and honestly, it's made hardly any difference Rust builds still take a ridiculously long time battery dies within an hour or so with a couple of agents running and I'm still not running any local AI models

@RhysSullivan says the guy who just posted this without a link

not enough people know that links in main tweets are fine again please stop adding spaces to your links, i want to be able to click through and view your thing

Anthropic didn’t buy Bun, they bought Jarred’s context window

@steveruizok It’s the axe effect, but you get nerds

Can't stop thinking about how if you buy one of the machines from ASML you get some guys too

Sonnet 4.6 is still the only practical daily driver if you're paying for your own tokens

lambda man humble bragging

@Vjeux Count me in as a sponsor

I wanna go in vacations again to build a Claude Code based engine that can play arbitrary video games. Claude plays pokemon needed a pokemon-specific harness and was extremely expensive and slow as it had to run LLM on every single turn. Instead I think we should have Claude write code to do: - the inner loop of analyze frames, figure out what to do and send inputs. - the outer loop of understanding the various sub parts of the games and building strategies to improve them. The inner loop should be hot reloadable so that when an update is done you don’t lose all the state (as the game is still playing) and have a way to yield back to the llm if it finds itself in a position where it is stuck. I’ve had success with Kaparthy’s research loop to build and improve the outer loop for various sub systems. I think this can work out. Good news is that there’s an endless amount of single player games on steam to experiment this on.

Little high

@RhysSullivan Planning on giving this a shot with just-bash / ai SDK tools soon

The vision I have for Executor is to be able to be able to replace how agents use CLIs and MCPs today with a significantly better primitive to let you call anything. I'm still figuring out the concise version of what Executor is, but for those that want the long explanation: Executor doesn't care what you add. MCP, OpenAPI, GraphQL, Custom sources, it just represents them as a tool name, input schema, output schema under the hood. This lets you then call them however you want. Today that's implemented as a code mode MCP, but it could also be called via an Executor CLI, or dropping in a native client. Also, because of this representation, they can be called via gen UI dashboards your agent writes, they can also be used in reusable workflows. Teams: You're able to set up all of your sources once and now every member of your team has access to them. It also supports per user credentials vs global credentials. Each member of your team shouldn't have to figure out how to set up your agentic environment, you shouldn't have to toggle on / off MCPs while working with your agents. Your agents should be able to access all of your company resources in a way that isn't scary. Sandboxing: Executor executes the tool calls in a JavaScript sandbox, the agent isn't able to access any of the secret values, those are done via proxy calls. Destructive Actions: Executor preserves the semantics of what it imports from i.e GET/DELETE for OpenAPI, destructiveHint for MCP, mutation for GraphQL. This then informs what the agent can auto run / not auto run, letting it work autonomously without removing you from the loop. It's all open source and built on the same SDK that's published to npm. Runs locally, will be self hostable in the cloud, there's so much to build and I'm really excited about it

@jarredsumner But how long does it take to compile?

99.8% of bun’s pre-existing test suite passes on Linux x64 glibc in the rust rewrite

Wow they gave up on Atlas

Omg long press share button in X for system share menu omg

Cursor now has a “burn through your $200 plan in an hour” button

Are people just texting during board meetings?

@2DoApp At this point 2Do is going to outlast Dropbox itself

Efforts to improve 2Do's innards, especially the impossible-sync-engine, have reaccelerated. This opens the door to more sync options and better interoperability in future. Still a lot to finish, but with my head down and moving forward 🙌. A small glimpse of the website refresh

I know you dimmed the context ring, @cursor_ai. I'm still going to stare at it all day every day, no matter how dim you make it.