I have been reading a lot about LLMs, MCP, Harnesses, etc., all the new AI stuff. I learn from first principles and try to nail down the basics. So far, one thing I understand about LLMs: they are stateless. You provide input tokens (your message), and they provide output tokens; it is like a black box. It can’t talk to the outside world on its own; that capability is added in a harness....
AX is Google’s new take on running agent workloads on Kubernetes: you declare a Task, a Workspace, a Gateway, and a Model as YAML, and it sandboxes the whole thing, wires up the workspace, fences the network, and lets you suspend and resume the agent like a checkpointed VM. It sits on top of Agent Substrate, Google’s sandboxed execution runtime, which is the part that actually does the suspend/resume and the gVisor isolation....
This post is entirely AI-generated, from a AI session. TL;DR: Clerk’s Chrome extension SDK needs syncHost pointed at your actual Frontend API domain, not your companion app’s domain. I had it pointed at the wrong one, so the SDK silently fell back to a cookie-based session mode that can never work inside an extension (the fetch layer sends credentials: "omit" on every request). Sign-in looked like it worked. It didn’t....
If you already use OpenCode but your models are exposed through an AI Gateway instead of the model provider’s public endpoint, the setup is straightforward once you know where OpenCode expects the provider URL to stop. This post walks through configuring OpenCode to use Anthropic-hosted models through an AI Gateway, the mistakes that commonly cause Not Found errors, and the small details that make the setup reliable. The goal We want OpenCode to:...
In this tutorial, we will learn how to run an N8N workflow using GitHub Actions. N8N is a powerful workflow automation tool that allows you to connect various services and automate tasks. Prerequisites Before we begin, ensure you have the following: A GitHub account with a repository where you want to set up the action Basic knowledge of GitHub Actions and N8N workflows A workflow created in N8N that you want to run How are we going to do it?...
TL;DR: I tried running n8n on Google Cloud Run without a managed database because I thought it would be cheaper. It “worked,” but it wasn’t straightforward: I had to bolt on automation for import/export state, fight Cloud Run restarts, keep CPU always allocated to keep a background backup loop alive, and deal with re-auth every few minutes when instances restarted. Fun experiment, not a recommendation. In this post, I’ll dive deep into how I made it work, what breaks, and why I’d pick PostgreSQL for n8n in anything beyond a toy project....
Introduction I have seen several documentation websites that have implemented AI chatbots (”Ask AI”) to allow users to ask questions about their documentation instead of manually searching. I believe this enhances user experience and saves time by allowing users to ask questions directly instead of manually searching through extensive texts. Large Language Models (LLMs) have enabled this use case of chatting with textual data. However, pre-trained LLMs typically lack up-to-date or specific knowledge about particular products....
Fluent Bit is a popular open-source log processor and forwarder that allows you to collect data from different sources, filter, and transform it before forwarding it to different destinations. In some cases, the data collected may contain sensitive information like passwords, credit card numbers, social security numbers, and other personal identifiable information (PII). To protect such information, you need to mask or obfuscate it before forwarding it to the destination. In this document, we will discuss how to mask sensitive information using Fluent Bit....
Hello everyone, This article is my take on what I have understood about the Kubernetes architecture. Instead of explaining the different components that the architecture comprises of & what each component does functionally. We will take a Kubernetes feature & break its implementation to understand how it interacts with Kubernetes system components. So yeah, some working knowledge of K8s basics such as creating deployments, and services is a plus. Before starting, I want to let you know that this article is written to the best of my knowledge & I’ll be updating it as I gain more insights on it....
In the previous article, we have learnt the journey of deployment creation in Kubernetes. Which helped us understand almost every aspect of the core system components of Kubernetes, but it missed the following components: Cloud Control Manager Kube Proxy To understand the above components, we will use the same technique of previous article where we will take a Kubernetes feature & break its implementation to understand how it interacts with the Kubernetes system components....