The Agentic Memory Trap
Lore is a vendor-neutral proxy that runs on your machine. Plain Markdown and an open SQLite database, run by a fair-source engine.
- infrastructure
- open standards
- data
- ownership
Blog
Lore notes, from the memory layer outward.
Product updates, architecture deep dives, and practical writing about persistent memory for AI coding agents.
The Semantic Linter: Automating reviews without losing touch
You can't write down your taste. But you keep working by it, and Lore learns it the way an apprentice would, then helps enforce it in code review.
- knowledge
- code review
- ci
- agents
Distill the knowledge, not the model
Satya Nadella named the Reverse Information Paradox. His fix over-solves it: your learning can stay yours in token space, without training a model.
- memory
- context management
- agents
How Lore remembers, forgets, and changes its mind
Most memory tools now ship a set of principles for how an agent should manage its context. Lore's principles work differently, because the layer enforces them instead of asking the agent to. Here are the rules Lore runs on.
- memory
- principles
- agents
Why memory is not enough
A long-term memory store remembers what you said last week. It can't manage the context window that's overflowing right now. Those are two different problems, and only one of them is getting solved.
- memory
- context management
- agents
Introducing the Lore Blog
A home for product notes, memory architecture deep dives, and engineering updates from Lore.
- memory
- agents