James Read's Code, Containers & Cloud blog

I’ve spent a fair bit of time on OliveTin documentation over the years — migrating to Antora, chasing broken links, adding search, and generally trying to make it usable for humans. That still matters. But recently I’ve been thinking about a second audience: LLMs and AI agents.

When someone (or something) asks “how do I configure OliveTin webhooks?” or “what’s the difference between shell and exec actions?”, the answer is probably already in the docs. The problem is getting there. HTML documentation sites are great in a browser — navigation, search, tabs, admonitions — but they’re awkward for an AI agent that wants a clean, text-first view of your content without scraping dozens of pages.

This is a brief snapshot post — mostly for future-me to look back on — about my early success with Hermes Agent. I’ve only been running it for 9 days, so treat everything here as first impressions, not a long-term review.

I tried OpenClaw in the past. It was interesting — I even mentioned it when writing about AI autonomy levels — but it never quite clicked for me. Hermes feels different so far: it seems to understand what I’m asking for more often, with less re-explaining and less “yes, but that’s not what I meant.”

One of the healthiest habits I’ve picked up over the years is to keep trying new tools — even, and especially, for things I use dozens of times a day. That applies whether you’re switching between commercial apps, trying a different free software project, or, in my case, maintaining your own fork of something you could just install from a package manager.

Getting attached to tools keeps you in the past. The terminal you patched in 2021 made sense in 2021. The editor you learned in 2011 made sense in 2011. But the ecosystem moves on, and if you’re not periodically asking “is there something better now?”, comfort slowly turns into friction you stop noticing.

Many people don’t have access to enough GPUs running in their own internal network to run LLMs, and at the same time, they want to be able to able to surface data to that LLM without putting the data publicly available online. n8n offers a nice solution to do this.

OliveTin has been getting a lot of security scrutiny this week - as there’s been a relatively huge surge of security advisories come for the first time. I imagine that the project has been listed in a security forum or similar, and AI tools (and human eyeballs) are being used to looking for security vulnerabilities. This is a good thing, and I welcome more security reports, that will be patched, and will make OliveTin better.

AI tooling for software engineering has moved quickly from fancy autocomplete to the verge of fully autonomous engineering. Most people can see that AI adds value. The real question is no longer whether to use AI, but how much autonomy to give it.

Capability alone doesn’t justify maximising autonomy everywhere. You have to match autonomy to risk. This post defines a simple scale of autonomy levels and where I draw my own lines.

I’ve been trying to wrap my head around NVidia’s portfolio to get a better understanding of the company. I’m very familiar with the consumer GPUs and parts of the data centre portfolio, but it’s taken me much longer to understand than expected! I kept coming up against marketing buzzwords and confusing naming - so this is a quick blog post to document my current understanding. I was particularly motivated to write this after reading about NVidia’s announcement in December 2025 - about reprioritising their DGX Cloud offering. I wanted to understand what that meant in context of their wider portfolio.

A quick snapshot of the cloud providers I’m still using in 2026—what I use, why, and a few recommendations that may be relevant to single developers, self hosting enthusiasts, and similar.

This is a post about making OliveTin more resilient in 2026, by adding betas, adding a Codeberg mirror, increasing Matrix usage, improving documentation, testing and tools. None of OliveTin’s principles are changing, it’s license isn’t changing, and it’s maintainer aren’t changing. Just making OliveTin stronger.

This article started life in October 2024 as a “Initial reaction to using Cursor.AI for a few hours” - as the provisional title suggests, I was trying out Cursor.AI for the first time - and it was not love at first sight! As an eager user of vim (and now, neovim more recently) I was curious to see how well this AI-powered code completion tool would integrate into my existing workflow.