Benjamin Eckel

This works in a clever way—by doing code generation against Figma’s Plugin API, so Claude is translating your app code into Figma functions. From my experience it makes it much more reliable!

Today’s Figma MCP update makes it one of the strongest integrations with Claude Code I’ve seen. You can now use Claude Code to design in Figma with the the full context of your design systems.

I'm a big believer in the idea that code is the ultimate "handle" for an agent to manipulate your app. Really excited to work on this with the team.

can't believe @wasm_io is less than a week away! @and_prf and I will be there for our session "The State of Zero-Dependency Wasm" with the latest from #wazero and #chicory 🚀

Major breakthrough tonight! After several weeks getting the C-based Prism Ruby parser hooked up to the Chicory WASM runtime (AOT compiled to JVM bytecode), I now have a local version of the gem that uses *zero* native libraries. Write once run anywhere indeed!

@TravLauren nsarchive2.gwu.edu/NSAEBB/NSAEBB1…

Has someone written a book on the theory behind and effectiveness of “astroturfing” ? Where the rich pay the mega poor to act as if the rich have people that agree with them to subvert democracy rules? And bots that do same thing online?

@evacchi the unholy trinity

@bhelx It's kinda worse, it's bots, redditors and 4chan users all impersonating each other

Quick life update, I'm no longer the CTO of Dylibso. I'm now working as an IC at @figma. Really excited to continue my work at a place that values the things I love: Wasm, extensibility, and building for creators. simst.im/articles/why-i…

Much love to the @CloudNativeFdn Helm team using Extism to power v4's new plugin system. helm.sh/docs/overview 🎉 Congrats on the new release!

@simonw @_philschmid It's a great solution for portable/sandboxed compute like this. But using Wasm still has a high DX friction for the average developer.

@_philschmid I still think WebAssembly should be the ideal solution for the sandbox here - it's the world's most widely tested sandbox after all

Agents executing their own code is inevitable; it's too powerful not to happen. But I’m stuck on the architecture. Do we run this locally on the user's device, or safely in the cloud? Local sandboxes are the dream. Offline-first, zero latency, native access to your data. But can be easily done wrong. Allow Ingress/Egress? Access to environment variables? Permissions to filesystems, libraries? Remote feels more secure, easier to get started but much higher setup per individuall to access personal their data.

I'm looking to hire someone to help me build KCL! It's a programming lang for 3D design that we're making at Zoo. You'd be working closely with me on the KCL interpreter, the CAD API it calls, and our mechanical engineers. Rust-heavy role. zoo.dev/jobs/461918700…

🤔 Have you ever wanted a plugin system for your project that can support a bajillion different languages? @cmgriffing has you covered in his talk from SquiggleConf 2024 on @extism! Check it out! 🔽 links below

🆕 Releasing the entire MCP track! We curated the best talks and must-know topics in the MCP world from 1H 2025: - @chu_onthis, Anthropic: MCP's Origins + Startups - John Welsh, Anthropic: shipping Remote MCPs - @altryne (W&B) + @bhelx (Dylibso): Observable .tools for MCP - @digitarald (VSCode): Hidden Capabilities of the Full MCP Spec - @samuelcolvin (Pydantic): MCP is all you need - @jancurn (Apify): the rise of the Agentic Economy - @zeeg (Sentry): MCP is not good yet! plus: - Den + @Jujujuliakasper (Microsoft): MCP Auth - @labeveryday: Agents with Amazon Nova + MCP! - @danmason (Stride): Telemedicine Case Study and more to come! youtube.com/playlist?list=… Enjoy!

I know we are all constantly trying to create specs, figuring out how to optimize cursor, claude code, and ship like no tomorrow But every time I use @extism I get "I am a real software engineer" vibes and it is so awesome

🚀Excited to share Quarkus quickjs4j, the new Quarkus extension by @EricWittmann that embeds the lightweight QuickJS engine in JVM & native builds! 🧩 It has never been that easy to run fully sandboxed JS inside your Java apps. Dive in 👉 quarkus.io/blog/quickjs4j #Java #Quarkus

When we created Chicory i theorized this would be possible, but had no idea someone would eventually do it! nice work @evacchi

@chu_onthis @AnthropicAI next on the MCP track: @altryne and @bhelx on the new standard for MCP Observability! youtube.com/watch?v=Lcqat4… check out observable . tools! Alex didn't top the judge outfit from last year but we like the @wandb swagger

@MichaelFor8032 @nilslice we're discussing this internally. as steve said, we're still trying to get them to approve our integration so we can't change scopes just this moment. But we can perhaps allow you to supply your own scopes.

@nilslice @bhelx @nilslice @bhelx I've submitted an issue on the mcp.run-servlets GH repo, but the google-mail servlet doesn't seem to be in that repo. The users_threads_modify tool is one of the claimed tools for your google-mail servlet but it doesn't work out of the box.

Don't let outdated access controls prevent you from shipping great AI integrations... Filter tools from MCP servers, and apply a consistent tool list across any client or agent! LIVE now on mcp (dot) run.

@pranay01 @aiDotEngineer @altryne @pranay01 signoz is great! you've made it so easy and pleasant to setup and experiment with.

🚀 Model Context Protocol (MCP) just met its new BFF: Observability 🚀 Was tuning into the @aiDotEngineer livestream (arguably the top LLM conf right now) when I suddenly spotted familiar purple traces on-screen—turns out @altryne & @bhelx picked SigNoz to share specific examples about MCP observability with the ai community ❤️ What’s MCP? Model Context Protocol—a spec for chaining model calls, tools, and function invocations into one coherent workflow. it's superpower? It captures the context that flows through every step of an LLM-powered app. Why observability matters: 1️⃣ LLMs are non-deterministic—latency and accuracy swing wildly with prompt tweaks. 2️⃣ Tool & RAG pipelines add hidden hops (vector DBs, HTTP fetches, markdown conversion…). 3️⃣ Without tracing, you’re left guessing where the 22 s disappeared. Thankful to the AI/LLM community for the love—and pumped to keep pushing the boundaries of what “observable AI” looks like! 🙌