OpenTeams

Running LLMs locally? The spec that actually drives your tokens/sec isn't VRAM or core count.
It's memory bandwidth.

Dillon Roach explains why, and what the numbers look like across consumer hardware.

Full breakdown on the OpenTeams blog:

What Hardware Do I Get!? | OpenTeams | AI you own Tech Lowdown with Dillon Roach

A skill, written as a simple instruction file with no code, no compilation, no deployment step. That's all it took for Adam Lewis to get a working Claude Code integration for time tracking running in about an hour.

The skill logged, edited, and deleted entries, using nothing more than a text file describing the CLI, his billing setup, and his preferences.

But it also exposes the limit of instruction-only systems. A skill can tell an agent what to do, but it can't enforce what the agent is actually allowed to access.

Adam breaks down when a skill is enough, when it stops being a security boundary, and how to reduce credential exposure with stricter controls.

Find out more on the OpenTeams engineering blog: https://na2.hubs.ly/H06vX540

From Skill to Agent: When a Text File Isn't Enough When does a Claude Code skill stop being enough? See why credential security pushes real workflows toward proper agent architectures.