01
Agentic Coding and Security
A four-layer risk model for coding agents: work ethic, mechanical gates, reversibility, and context hygiene.
Open articleThe English Garden has started growing.
The first translated Garden texts are now live in English. The companion glossary has moved over too, so the English Garden is now a proper starting point.
02
Glossary
A growing companion sheet for terms like session anchor, pre-push hook, dependency audit, rollback, and prompt injection.
Open glossary03
Lethal Trifecta and Prompt Injection
Why untrusted input, sensitive data, and tools with impact become a real problem together, and what the Rule of Two does about it.
Open article:-) The Garden is a growing space for working notes and thinking in public.