Agent Engineering Foundations
Harness, Gateway, context engineering, evaluation, framework selection. How to make an Agent run reliably.
This module answers another question: how do you make an Agent run reliably in production?
The core idea in one line — the model is the driver, the engineering infrastructure is the car. Once model capability is commoditized, the moat moves up to the Harness, context engineering, and the data flywheel. Starting from the Harness overview, this module covers context engineering (managing the scarcest resource), the Gateway (the production chassis), framework selection (don't default to LangGraph), the evaluation system (don't iterate on gut feel), and multi-Agent platforms (evolve by bottleneck, not by imagination).
Lecture order: Lecture 0 (Harness) is the overview — fix the car before swapping the driver — then five lectures each map to one engineering subsystem, all following "what breaks without it → framework → how to ship it."
For: engineers and indie developers building Agent applications, and PMs who want to fill in the "engineering foundations" mental model.
Posts
Why a Stronger Model Won't Save Your Agent — Fix the Harness First
The model is the driver, the Harness is the car. If it can't deliver, it's usually the car's fault, not the driver's.
B · 1 Foundations · Post 1Why a Mismanaged Context Window Makes Even the Smartest Agent Dumb
Context is an Agent's scarcest resource—treat it as a system to be engineered, using a seven-dimension lens and a composition audit.
B · 2 Foundations · Post 2Why the First Thing to Break in Production Isn't the Model — It's the Plumbing
The Agent Gateway — ingress, routing, persistence, reliable delivery, and concurrency that let an Agent truly live in production.
B · 3 Foundations · Post 3Why You Should Pick an Agent Framework Based on Need, Not Default to LangGraph
A decision tree and trade-off table that make clear when to reach for LangGraph / Checkpoint / Map-Reduce / Multi-Agent—and when not to.
B · 4 Foundations · Post 4Why an Agent Without an Eval System Can Only Iterate on Gut Feel
A thorough walkthrough of the four pieces — Agent eval dimensions, building test sets, input-quality tiered routing, and model-routing selection.
B · 5 Foundations · Post 5Why a Multi-Agent Platform Can't Be Built in One Shot — It Has to Grow in Four Stages
The four-stage evolution from a single Agent to a multi-Agent platform, plus the three new costs it introduces — handoff, sharing, and aggregation.