GenAI | Fresh/Brewed

A2A

Bring me my Llama

Ever since I heard about the Agent to Agent (A2A) protocol, I’ve been curious what it was and how it might be useful. The general idea of A2A is that it’s a standard that Agentic workloads can leverage to discover and talk to each other using “cards” that advertise capabilities. We will look at two A2A examples, how they work and test them. Then we’ll try and recreate them with python containerize servers with REST endpoints.

Browser Use

Agentic web crawlers for fun and profit

I had a bookmark (not sure where I saw it) for Browser Use. It’s a geared for agentic work, of course, but the idea is ephemeral and scheduled web browser related flows that are easy to script with NodeJS or python. It has a free tier so there is no reason not to give it a go and see what it can do. Setup We can signup here with username/password or a federated IdP. I’ll do the latter with google

Creating and using AI Skills

this can save you money

I’ve had AI Skills on the mind lately and realized I had neglected updating a few of my own and it was high time to handle that. In this article we’ll cover installing and using some common public skills, like Anthropic’s skills and specifically their “skill creator” skill. We’ll use that in Google Antigravity and Copilot to cleanup an older Python App skill I have. We’ll then test using the Python App skill in Cline, a local CLI Harness with a 12b gemma4 model running in Ollama on a laptop with a 12Gb 3070 card in it.

Nextchat Librechat and Mistral AI

one is worth using

I had a couple of sites bookmarked to circle back on: Nextchat.club and LibreChat. I’m not certain what they really are, just that they were on an AI startup leader board recently. The other AI offering I haven’t visited in some time is Mistral. I liked the early Mistral models for coding. I also like that it’s based in France. Nextchat.club We can sign-up at nextchat.club/login: in the Application area we can pick a persona. Here I’ll pick a resume writer

A 1980s Airline Repair guide may be a cure for AI Slop

How a 30 year old reduced English standard applies to LLM tech writing today

In the last few days I’ve been rather fixated on this YouTube video called “The cure for AI slop is a 1986 aircraft manual”. If you like more of a narrative, pause here and watch the video. First, a bit of context… Simplified Technical English has its starts going all the way back to the 1930s with “Basic English” and later the 1970s with CFE, Caterpillar Technical English and Kodak International Service Language (KISL). Per the CFE documentation:

Fresh/Brewed

A2A