AI & Agents
AI & Agents
A practical reading path through the engineering around modern AI systems.
These articles focus less on model hype and more on the surrounding system: context, tools, agent loops, local models, verification, security, and the workflows that make AI useful in real work.
Start with the foundations
LLM Strengths and Limitations: A Practical Framework Understand what language models are good at, where they fail, and how to design around those limits.
AI Application Architecture: LLM + Memory + Tools See why a useful AI application needs more than a model and a prompt.
MCP, Skills, Agents, Rulesets, and Hooks A map of the components that make up a governed AI development environment.
Understand coding agents
How an LLM Coding Agent Actually Builds Software The model, context builder, tool runtime, agent loop, and verification layer explained from the inside.
Giving Hermes Durable Codebase Context with GitNexus, LSP, and AGENTS.md How different kinds of codebase knowledge can have different owners instead of forcing grep to do everything.
The Hard Part of AI PR Review Is Not Reading the Diff Why review agents need evidence, repository context, and a bias-resistant verification loop.
Build with local models and safer workflows
Using LM Studio and Gemma as a Local Engine for Coding Agents A local-model setup for experimenting with coding-agent workflows.
Orchestrating Hermes with Local Gemma Cheap routing, delegation, explicit boundaries, and safe Rails workflows.
Golden Rules for Cheaper, Safer LLM Agents When to use an LLM, a workflow, a script, a command, or a scheduled job.
Don’t Leave Good AI Workflows in Chat A practical pattern for promoting repeated work into cheaper and more reliable layers.
Continue exploring
This page will evolve as new articles add tested patterns, implementation notes, and lessons from real systems.