I think this is the fastest/easiest way to both deploy and consume a remote MCP server, courtesy of @OpenAIDevs and @deno_land with Deno Deploy. ~ 30 lines of code for a single file server ~ 18 lines of code to have an OAI model use it Tough to beat that. Some details 🧵

This is something we're working very hard on at @StainlessAPI given our proximity to API specs and SDK implementations. We feel like we can get really, really good at helping LLMs write integration code.

Most API docs only show HTTP endpoints, but developers write code with SDKs. We're doing things differently with the Stainless Docs Platform, now available in early access. Stainless Docs are SDK-native. They update automatically when your API changes. Devs see your docs in the language of their choice with method signatures, types, and accurate code snippets.

42g of protein in one of these, crazy. And it does legit just taste like a bottle of flavored milk.

This was published one year ago. Everything is right.

Terminal 2 at MSP is wild. All the greatest hits of class trips, dads with tickets printed on 8.5x11 inch paper, and "oh sorry I didn't realize I couldn't take this gallon of 2% milk in my carry-on".

Love to see this role opening up! Can confirm it was very awesome to influence such a critical API by even 0.0001%, so this one sounds really cool.

Over the past few months at @dedaluslabs, we’ve noticed how many talented founders from around the world are being held back by one thing: location. There is no better place in the world to build an AI startup than San Francisco. SF is home to all the major AI labs, top talent, world-class investors, and it is where the future is being built every day. That’s why, this December, we’re giving builders everywhere a chance to break in. We are taking over @_TheResidency and launching Break In—a month-long hacker house program in the heart of San Francisco. If you’re not from SF, and you’re building an AI startup, this is your chance to join us. Build your startup on top of Dedalus’ Agents SDK or MCP deployment infrastructure and we’ll cover your stay (housing, lunch, and co-working space). As a member of Break In, you’ll get access to all the best parts of SF. You’ll be introduced to top founders, mentors, and investors. You’ll get to skip the line at the biggest tech events. Plus, you’ll receive free API credits and expanded access to the best-in-class AI stack through our partner network of top AI infrastructure companies. Applications open October 27th and close November 10th. We will be accepting applications on a rolling basis. Final acceptances will go out November 14th to ensure international founders have enough time to secure visas. Ready to break in? Follow us on Twitter/X for more updates and visit the link in the comments to apply on October 27th!

Hi @FredKSchott and @astrodotbuild! Stainless has a big product coming out soon that has Astro at the core. I have a partnership request in, but is there anyone on the DevRel or marketing side I can chat with as well? Thanks!

I'm not a protest type. But the casual dismissal of #NoKings as mere lefty cope, or the active smeering it as anti-American, is, well, anti-American. Protest is a great American tradition. The ongoing assault on the Constitutional order deserves, at least, this, from all of us.

As a newly minted head of marketing, I have one major goal - to create a launch video where the human participants appear, in actual fact, to be humans who are not hostages being held at gunpoint and forced to talk at a camera.

Awwwww a wild Christina Huang appeared

luckily they rescheduled EnterpriseValueDay or I would have made the same mistake.

Come play ArcadeGPT today at @OpenAIDevs DevDay! In my sophomore conference arcade game effort, I made 7 games inspired by arcade classics that you can remix in weird ways with GPT-5. s/o hardware engineer Daisy

Live from DevDay with @sama. 10am PT. openai.com/live

If your MCP server has dozens of tools, it’s probably built wrong. You need tools that are specific and clear for each use case—but you also can’t have too many. This creates an almost impossible tradeoff that most companies don’t know how to solve. That’s why I interviewed my friend Alex Rattray (@RattrayAlex), the founder and CEO of @StainlessAPI. Stainless builds APIs, SDKs, and MCP servers for companies like @OpenAI and @AnthropicAI. Alex has spent years mastering how to make software talk to software, and he came on the show to share what he knows. I had him on @every’s AI & I to talk about MCP and the future of the AI-native internet. We get into: • Design MCP servers to be lean and precise. Alex’s best practices for building reliable MCP servers start with keeping the toolset small, giving each tool a precise name and description, and minimizing the inputs and outputs the model has to handle. At Stainless, they also often add a JSON filter on top to strip out unnecessary data. • Make complex APIs manageable with dynamic mode. To solve the problem of how an AI figures out which tool to use in larger APIs, Stainless switches to “dynamic mode,” where the model gets only three tools: List the endpoints, pick one and learn about it, and then execute it. • MCP servers as business copilots. At Stainless, Alex uses MCP servers to connect tools like @NotionHQ and @HubSpot, so he can ask questions like, “Which customers signed up last week?” The system queries multiple databases and returns a summary that would’ve otherwise taken multiple logins and searches. • Create a “brain” for your company with Claude Code. Alex built a shared company brain at Stainless by keeping Claude Code running on his system and asking it to save useful inputs—like customer feedback and SQL queries—into GitHub. Over time, this creates a curated archive his team can query easily. • The future of MCP is code execution. Instead of giving models hundreds of tools, Alex believes the most powerful setup will be a simple code execution tool and a doc search tool. The AI writes code against an API’s SDK, runs it on a server, and checks the docs when it gets stuck. This is a must-watch for anyone who wants to understand MCP—and learn how to use them as a competitive edge. Watch below! Timestamps: Introduction: 00:01:14 Why Alex likes running barefoot: 00:02:54 APIs and MCP, the connectors of the new internet: 00:05:09 Why MCP servers are hard to get right: 00:10:53 Design principles for reliable MCP servers: 00:20:07 Scaling MCP servers for large APIs: 00:23:50 Using MCP for business ops at Stainless: 00:25:14 Building a company brain with Claude Code: 00:28:12 Where MCP goes from here: 00:33:59 Alex’s take on the security model for MCP: 00:41:10

Yeah, it's definitely been the case for me (using Cursor specifically) that 4.5 Sonnet's speed advantage over gpt-5 makes it useful for simpler prompts, but if I need something difficult done properly I still need gpt-5 with high juice.

Feels right to me too. 4.5 has a very favorable speed/intelligence ratio, gpt-5 better at harder problems.

On a happier note - my fellow game dev and I are making progress

Don't be tricked into thinking that political violence is exclusively (or even primarily) targeted at conservatives.

Donald Trump can shut right the fuck up about encouraging domestic terrorism after inciting a mob to attack the Capitol. Notably all his hand wringing about political violence exclusively mentions violence against his allies, not his political enemies.