[Dev Catch Up # 79] - ChatGPT supports MCP Servers, Rest API Essentials, Gonzo, Authentication 101, Claude Code in Zed, Claude can create ppt & excel, Why LLMs hallucinate, EmbeddingGemma and more!
Bringing devs up to speed on the latest dev news from the trends including, a bunch of exciting developments and articles
Welcome to the 79th edition of DevShorts, Dev Catch Up!
For those who joined recently or are reading Dev Catch Up for the first time, I write about developer stories and open source, partly based on my work and experience interacting with people all over the globe.
Thanks for reading Dev Shorts! Subscribe for free to receive new posts and support my work.
Some recent issues from Dev Catch up:
Join 8000+ developers to hear stories from Open source and technology.
Must Read
Anthropic has released a new update. Claude can now create and edit files including Excel spreadsheets, documents, PowerPoint slide decks, and PDFs directly in Claude.ai and in the desktop app. Check Anthropic’s post for more details.
OpenAI has finally added MCP support in ChatGPT. You can now connect remote MCP servers directly with ChatGPT. This feature is now available for Pro and Plus accounts on the web. Check this page to know how to add MCP Servers with ChatGPT.
Authentication is the first gate in any system. It has many forms like Basic, Bearer, OAuth2, JWT, and SSO. This guide explains each one with clear examples. Check this Substack post to know more details on how authentication works.
Server-Sent Events (SSE) is a simple way to send real-time updates from server to client. It works over a single HTTP connection and is well-suited for live dashboards, notifications, and streaming data. Read this Substack post to learn what SSE is and how it works.
OSS Highlight of the Week
This week we are featuring Gonzo, a real-time log analysis tool for the terminal. It helps you analyze log streams with charts, AI-powered insights, and advanced filtering. It also has an interactive dashboard for deeper drill-down. Check the GitHub repo to explore more.
Good to know
If you are building REST APIs, you should definitely know about HTTP methods, status codes, and resource modeling. This Substack post covers how to design clean, consistent REST APIs. Check it to learn REST API proven best practices.
Transformers are hard to understand because of all the complex math happening inside. This interactive demo makes it simple by showing how attention works with just 19 tokens. So, you can see exactly how the model processes information. Check this demo if you want to understand how LLMs actually work.
I came across MediaBunny, a JavaScript library for handling media files directly in the browser. It lets you read, write, and convert video and audio files without needing server-side processing. Check this GitHub repo if you want to process media files on the client side.
OpenAI shared a post on why language models hallucinate. Current evals push models to answer even when they are unsure. When models say, "I don't know," eval systems give no credit. This leads to confident but wrong answers. Check this post to learn more details on this.
While exploring agentic frameworks, I got to know about Parlant. It helps you build AI agents that reliably follow instructions and stay aligned with your protocols. Check this page to know more about Parlant.
Notable FYIs
Google has released EmbeddingGemma, a lightweight open embedding model built for search and retrieval tasks. It is small enough to run on-device and works offline, making it useful for building RAG apps anywhere. Read Google’s post to know more details.
Alibaba released Qwen3-Max-Preview. It is their biggest model with 1 trillion parameters. Tests show it works better for conversations and complex tasks. Check the announcement for launch details and how to access it.
Zed Code Editor now supports Claude Code through the new Agent Client Protocol (ACP). This protocol allows any agent to connect to Zed and other editors. Check their post to see what they have announced.
If you are using OpenAI’s legacy Completions or Chat Completions APIs, there is now a migration pack to help you move to the unified Responses API. It comes with a one-liner script and Codex CLI support to make upgrades easier. Check the GitHub repo for more details.
That’s it from us with this edition. We hope you are going away with a ton of new information. Lastly, share this newsletter with your colleagues and pals if you find it valuable. A subscription to the newsletter will be awesome if you are reading it for the first time.