Blog
Notes on agent memory
Causal memory, decision lineage, and making autonomous agents auditable.
2026-08-22
The Stale-Memory Problem
Your agent acts on facts it already updated. MADB 0.2.13 adds current-truth recall — mark a memory superseded and recall returns what’s current, with as_of for history.
Read →2026-07-29
The 2030 Memory Problem
Agents that build agents won’t be limited by reasoning — they’ll be limited by memory. The case for causal, auditable memory, and how MADB solves it today.
Read →2026-06-30
What Is Causal Memory for AI Agents?
The idea behind caused_by edges and trace_cause — and why similarity-based memory isn’t enough.
Read →2026-06-30
How to Give Claude Code Persistent Memory
A two-minute walkthrough: install MADB, store a memory, and recall it in a fresh session.
Read →2026-06-30
Why AI Agents Need Decision Lineage
Autonomous agents make real decisions. Without lineage, no one can explain or defend them.
Read →