Actions (Forgejo actions)

So I finally got to setting up some Forgejo runners and deployment actions. Very worthwhile and much lighter than I thought it would be.

It started with adding Renovate which gave me pages of homework. Getting that merge commit and deploy structure in place was a huge bonus. Now continuing to knock out my todos and get everything cycled with proper versioning and updates.

(aside):

• this line is to make sure this thing actually deployed this time. 🤞
• Git sub modules are a pain.
• A real pain

Forgejo test. Migrate.

In order to keep moving to an unfettered homelab, started migrating Gitea to the Forgejo fork. I missed the opportunity to do the easy migration. But after a bit of Postgres and Longhorn volume finagling I finally seem to have gotten things stable. So far, it has been a pretty easy switch just for some reason still missing a few repos that I need to manually migrate. Claude’s script seemed to miss a few.

Read more…

I saw a post for Digital Independence day on Mastodon. I spent a lot of time moving my data to the cloud, into the safe hands of the cloud custodians and infrastructure. I remember how cool and liberating that felt. The current ecosystem shows how fragile that plan was now. It’s unfortunate to see how all the titans fell in unison and now I feel nothing is protected - or protected under terms that can be changed at a whim. Now I’m clawing it all back and it’s just as liberating. Trust no one.

Read more…

• Perplexica latest updates migrated codebase to postgresql support to get away from local SQLite file.

Coda:

So the developer anonymously came forward to throw his bot under the bus. It’s striking with the lack of accountability. When humans kick off automated processes, it’s not the code’s fault, the human holds the accountability. The same goes for AI and AI agents.

I am not against using AI and agentic AI as a tool. But when 💩 happens: improper use, poor coding, theft, whatever… it absolutely flows downhill and lands on the human.

Read more…

Things I love (again):

• Personal media
  • Heavy metal.
  • Not so heavy metal.
  • Old songs I still have the cassettes to.
  • The library - Libby, CloudLibrary
  • Audiobooks
    • Life of Chuck
    • Stephen King
    • Murderbot
• Podcasts
• Linux, the homelab, personal cloud
  • Nextcloud
  • FastMail
  • Audiobookshelf
  • Jellyfin
  • Fediverse
• Mastodon and the Fediverse
• Responsible AI

AI Agents, Fleets, and Their Autonomous Capabilities

Introduction

AI agents represent a significant evolution in artificial intelligence, moving beyond traditional chatbots to systems capable of autonomous, goal-directed action[8]. This executive summary addresses your core questions about what these systems are, how they differ from earlier agent-based computing paradigms, and their current real-world applications—particularly through Anthropic’s Claude implementation.

What Are AI Agents?

Core Definition and Capabilities

AI agents are intelligent systems distinguished by their ability to operate autonomously in complex environments[3]. Unlike traditional conversational AI, they can process multimodal information including text, voice, video, audio, and code simultaneously[5]. More importantly, they possess the ability to converse, reason, learn, and make independent decisions[5].

Read more…

AI Agents, Fleets, and Their Real-World Applications

Introduction

AI agents represent a significant evolution in how artificial intelligence systems operate within business and technical environments. Unlike traditional chatbots that respond to queries, AI agents are designed to autonomously set goals, make decisions, and execute complex tasks with minimal human intervention[1]. This executive summary provides a technical overview of what AI agents are, their autonomous capabilities, their organizational potential through agent fleets, and their current real-world applications—particularly through Anthropic’s Claude implementations that have recently gained prominence.

Read more…

I started many blogs in my career. I imagine this space will be the same sporadic updates I normally come up with. In particilar, I’m thinking I could try to document some of my #homelab experiences. Tips and tricks, gotchas, and the like. I’ll probably focus on practical guides and troubleshooting notes that might help others navigating similar setups.

Read more…