[Dev Catch Up # 80] - Google's Agent Payments Protocol (AP2), Clean Code Tips, API Security, Gemini in Chrome, MCP Containers, Orchids - The AI Full Stack Engineer, AI Toolkit for VSCode & much 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 80th 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
Google has announced Agent Payments Protocol (AP2). It is an open standard for agent-driven payments. It addresses how agents can make payments without human supervision. Read Google’s blog for more details on how AP2 brings trust into agent-driven commerce.
Orchids enters as a new player in AI development tools. It positions itself as the first AI full-stack engineer. It claims it is capable of implementing frontend, backend, auth, database, and payments without relying on third-party integrations. Check the Twitter post for the announcement.
Clean code is the base of maintainable software. It’s not just about passing tests, but about making code easy for others to read and use. Read this Substack post that covers 10 simple tips for writing clean code using clear names and keeping functions small and focused and more.
APIs are the backbone of any app, and their security is a must. Weak security can leak data and make systems unsafe. This Substack post shares 9 key practices like HTTPS, authentication, rate limits, and input checks to help protect your APIs from common threats.
OSS Highlight of the Week
This week we're featuring doxx. A terminal document viewer for Word files. It helps to view, search, and export .docx documents without leaving the command line. It supports multiple export formats like Markdown, CSV, and JSON. It is perfect for developers who want to handle Word documents in the terminal. Check the GitHub repo for more details.
Good to know
Caching choices in an application shouldn’t be random. It should be a thoughtful choice based on questions like how often the data is used, how much it costs to fetch, and how stable it is. Here is the substack post that talks about a seven-question framework for making better caching decisions.
I came across Genkit, Firebase’s open-source framework for building AI apps. It works with models from multiple providers like Google, OpenAI, Anthropic, and Ollama. Check the GitHub repo if you want to explore Genkit.
Every developer should know the basics of system design, from caching to sharding. These concepts are key to building scalable apps that handle millions of users. This Substack post explains 11 system design topics like scalability, latency, consistency, and availability with clear examples.
Anthropic shared a postmortem on three recent issues that affected Claude’s responses. They clarified that Claude’s quality was never reduced due to high demand. The blog also explains what went wrong and the fixes they put in place. Check Anthropic’s post for details.
Getting consistent results from LLMs isn’t always easy, even if you set temperature to 0. Different batch sizes can change the math and lead to different outputs for the same input. This post from Thinking Machines explains why this happens and how to fix it for more reliable LLM inference.
Notable FYIs
Google is bringing Gemini right into Chrome. You can now get summaries, clarifications, answers, and comparisons directly in your browser using the context of your open tabs. Read Google’s post to see how Gemini in Chrome makes browsing more interactive and helpful.
If you’re using MCP servers, check out the MCP Containers repo. It has hundreds of containerized MCP servers that get updated daily. You can just pull the Docker Image to use any MCP Server.
ATLAS is a new transformer-like architecture that replaces attention layers with a trainable memory module. It can handle up to 10 million tokens, far more than today’s models. In tests, it reached 80% accuracy on long-context tasks, beating other models of the same size. Check the DeepLearning AI article for details.
VS Code now has an AI Toolkit extension. This extension turns VSCode into a complete AI development environment. It comes with key features like a model catalog, evaluation tools, and an agent builder. Read the VS Code docs to learn more about this toolkit.
Anthropic shared a guide on building better tools for LLM agents with the Model Context Protocol (MCP). It shows how to make quick prototypes and test them. The post also gives tips on choosing the right tools, returning useful context, and making responses efficient. Check Anthropic’s blog for the full guide.
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.