homelab 28
- Putting the Lab in a Chat Window: Agent Skills, Flows, and CI/CD for AnythingLLM
- Helen Gets a Command Surface
- Seven in the Morning: What Helen Says, and Why No Model Writes It
- A Refusal a Model Can't Be Talked Out Of
- How Helen Came to Be: Four Voices and a State Machine
- The Coach That Doesn't Use a Model
- Looking Outward: A Small Data Lake for Public Sources
- Two Stages and a Refusal: Routing Inside the Memory
- The Index That Was Quietly Empty
- Teaching the Lab to Know Me: Journals Into Structured Data
- Mostly Arithmetic: Classifying Intent Without Training Anything
- Higher Brain Monitor: A Physical UX for the Home AI Lab
- Where Does This Sentence Go? The Gateway Problem
- A Dropped Log Is Worse Than an Ugly One
- From RAG to a Knowledge Factory: An LLM-Owned Wiki That Compounds
- From Wake Word to Soundbar: The Whole Voice Pipeline
- An Operator Console, Not a Dashboard
- A Console Cyberdeck for Azure and Agentic Development
- Body Language: Giving the Lab a Face on the Wall
- Two Platforms On Purpose: systemd, k3s, and Refusing to Migrate
- Zero Redundancy: Backing Up a Striped Array I Chose On Purpose
- Is Any of This Even Running? Building the Lab's Single Pane of Glass
- 1.7 Tokens Per Second: How the Lab Got Its GPUs
- Load-Bearing Docs: How a Homelab Learns to Stop Lying to Itself
- Eating my own dogfood: ADR for the UI plane of my AI Lab
- CLI vs Open WebUI: Why Context Actually Matters When Running Local LLMs
- Homelab Seismic AI with Slack Alerts
- Earthquake Data, .NET, and Local LLM Reasoning