Saskia Teichmann
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.
Keyword
Context Engineering
Articles, workshop reports, and analyses on the topic of „Context Engineering“—drawn directly from my work with digital systems, e-commerce, and AI.
6 posts
More Articles on Context Engineering
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 …
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.
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.
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.
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.
News from the Studio
New posts via email.
Whenever a new workshop report or guide is published, you'll receive a brief email with a link. No set schedule, no ads.
Prefer to read via RSS