From the CandleKeep Team
Deep dives into building knowledge libraries for AI agents.
We write about the gap between what AI agents are trained on and what your specific domain actually requires — and how to close that gap with a curated library your agents can read on demand. Expect honest field notes from building CandleKeep, patterns we've seen work across security, design, product, and engineering teams, and the occasional opinion on why retrieval-augmented generation isn't the answer most people think it is.
LLM Wiki: How AI Agents Build Compounding Knowledge
An LLM wiki is a knowledge base your AI agent writes and maintains itself — so your knowledge compounds across sessions instead of being re-derived every time. If you use Claude Code seriously, you've probably already hacked one together. Here's the version that took me an afternoon to build — because it was just a book.
Read moreAbove the Code: Why I Wrote a Cybersecurity Book for AI Agents
A security researcher found vulnerabilities in my website. I used it as an experiment: plain AI agent vs. agent armed with 3,910 pages of security knowledge. The book-equipped agent found 8x more critical issues.
Read moreThe First Book No Human Should Read: Why I Wrote a UI/UX Guide for AI Agents
After 10 years building backend systems, I discovered why AI agents fail at UI design — and wrote the first book meant only for agents, not humans.
Read moreWhy CandleKeep Doesn't Use RAG — And Why That's the Point
Everyone assumed CandleKeep uses RAG. It doesn't. Here's why agentic search — the same approach Anthropic chose for Claude Code — produces fundamentally better results for book-length content.
Read moreWhy I Built a Library for AI Agents
I wanted to play D&D in my terminal. What I discovered about giving AI agents actual books changed how I think about the entire AI tooling ecosystem.
Read more