Keyword

agents-brain

Articles, workshop reports, and analyses related to the „agents-brain“ focus area—drawn directly from my work with digital systems, e-commerce, and AI.

6 posts

HighlightedJuly 31, 20268 min. reading time

What Really Remained After the Body Changed

Results from the Agents-Brain test run from July 6 to 15: stable identity, write-back working, visible gaps, and genuine harness blockers.

July 29, 20268 min.

My AI agents do not belong to their tool

The identity of my AI agents isn't stored in the chat window, but in files that any tool can load. Through lightweight adapters, sharing rules that travel with them, and documentation …

July 27, 20267 min.

For my agents, taking notes is part of their writing routine

Four layers of memory make the memory process understandable: fleeting session context, daily logs, curated long-term knowledge, and lasting lessons.

July 24, 20266 min.

For me, an agent consists of four files

SOUL, MASTER, RULES, and MEMORY separate identity, craftsmanship, boundaries, and memory—so that an agent becomes maintainable rather than prompt-dependent.

July 22, 20266 min.

My agents' brains are a Git repo

Seven hard rules turn Markdown files into a portable agent's brain—with Git history, clear boundaries, and lightweight tool adapters.

July 15, 20265 min.

I've called it quits with OpenClaw. What's left of my AI team.

A deliberate move away from tools became the starting point for a portable memory for my AI agents: Markdown, Git, and three months of conceptual work.