The Right 300 Tokens Beat 100k Noisy Ones

Abstract

Your agent has 100k tokens of context. It still forgets what you told it two messages ago. Prompt engineering taught us to craft the perfect instruction. Context engineering asks a different question: what does your model need to see and what should it never see at all? It’s the shift from writing prompts to designing context. In this talk, we dissect four antipatterns killing your agents and the architectural fixes that actually work: - The Stuffed Prompt – You crammed everything upfront and hoped for the best. Static context doesn’t scale. We explore dynamic loading and context refinement: fetching what’s needed when it’s needed. - The Wrong Tool for the Job – You picked one retrieval method and used it everywhere. RAG isn’t always the answer. We break down when embeddings help, when versioned doc artifacts give you precision, and when a simple lookup beats both. - The Goldfish Agent – Your agent forgets everything between sessions. We explore short-term and long-term memory, what to persist, what to summarize, and when to let go. - The Vibes Eval – You shipped because it “felt right.” We build eval strategies that prove your context choices work or expose the tokens you’re wasting. We use a coding agent to explain these patterns so you learn how they work under the hood – but everything also applies to AI agents in general.