Module 1 · Harnesses · Lesson 1

Choosing a harness

What an agent harness gives you beyond the model call, how Claude Code, the Claude Agent SDK, Claude Managed Agents and eve differ, and how to pick one for a production agent.

60 min 1 video 2 exercises 7-question quiz Answers

Setup for the exercises:Install Claude Code Node.js, pnpm and TypeScript

By the end of this lesson you can

List the responsibilities of an agent harness (loop, tools, context management, sessions, control surfaces, runtime) and say which ones you would otherwise build yourself
Compare Claude Code, the Claude Agent SDK, Claude Managed Agents and eve on who runs the loop, where state lives, model and billing, and durability
Apply a weighted decision matrix to pick a harness for a specific agent, and identify the deal-breaker criteria that override the score
Map the same concept (tools, skills, subagents, approvals, hooks, sessions) across the Claude stack and eve

Picture a fictional conveyancing firm that wants a dozen agents: one that reviews official copies, one that drafts pre-contract enquiries, one that chases replies, one that prepares the report on title. Each needs a loop that calls a model, runs tools, keeps context under control, remembers the conversation and lets a solicitor stop it before anything leaves the firm. You could write that loop yourself in an afternoon. Making it survive a redeploy halfway through a three-day wait for the seller's solicitor, and getting permissions and audit right, takes months. That is what a harness is for, and choosing one is the first architectural decision for every agent you build.

Module 1 · Harnesses · Lesson 1

Choosing a harness

What an agent harness gives you beyond the model call, how Claude Code, the Claude Agent SDK, Claude Managed Agents and eve differ, and how to pick one for a production agent.

60 min 1 video 2 exercises 7-question quiz Answers

Setup for the exercises:Install Claude Code Node.js, pnpm and TypeScript

By the end of this lesson you can

List the responsibilities of an agent harness (loop, tools, context management, sessions, control surfaces, runtime) and say which ones you would otherwise build yourself
Compare Claude Code, the Claude Agent SDK, Claude Managed Agents and eve on who runs the loop, where state lives, model and billing, and durability
Apply a weighted decision matrix to pick a harness for a specific agent, and identify the deal-breaker criteria that override the score
Map the same concept (tools, skills, subagents, approvals, hooks, sessions) across the Claude stack and eve

Free · sign in to start

Sign in to keep reading.

Free, no card. We use your account to save progress, quiz scores, bookmarks and occasionally send updates from our team.

9 more sections
2 hands-on exercises
A 7-question quiz
Progress, quiz scores and bookmarks