Skip to content
Sign in

Courses / Building with agents

Building with agents

An agent is a model, a list of tools and a loop. This course takes each piece apart, then shows you how to set one up and run it well.

Start here is for anyone who has used ChatGPT or Claude a few times. The modules after it are for people who build with agents: developers, technical founders and teams.

0 of 29 lessons out

Watch
A short film, five to seven minutes.
Read
The written lesson under it, with sources and dates.
Keep
One free download: a card, a template or working code.

Start here · For anyone · 4 lessons

No code needed.

  1. S1What a token isKeep: Tokens on one pageComing soon
  2. S2Setting up ClaudeKeep: A setup checklistComing soon
  3. S3How to askKeep: Ten starter promptsComing soon
  4. S4Your first skillKeep: Three paste-in skillsComing soon

Module 1 · How an agent works · 3 lessons

  1. 1.1What an LLM is, and the agent loop under the hoodKeep: The loop in 50 lines of PythonFirst up
  2. 1.2The core tools, and how a task flows through themKeep: The four tools cardComing soon
  3. 1.3How coding agents structure system prompts and toolsKeep: An annotated tool definitionComing soon

Module 2 · Asking and planning · 2 lessons

  1. 2.1Prompting techniques, and when each appliesKeep: Which technique, whenComing soon
  2. 2.2Research, plan, implementKeep: A spec templateComing soon

Module 3 · MCP · 2 lessons

  1. 3.1MCP fundamentals: servers, clients, toolsKeep: An MCP cheat sheetComing soon
  2. 3.2Designing tools an agent can use wellKeep: A starter MCP serverComing soon

Module 4 · Skills and the command line · 3 lessons

  1. 4.1What skills areKeep: A skill template and three skillsComing soon
  2. 4.2Web skills: reaching beyond the repoKeep: A web-research skillComing soon
  3. 4.3Working well from the command lineKeep: The command-line habits cardComing soon

Module 5 · Setting up the harness · 3 lessons

  1. 5.1CLAUDE.md and AGENTS.md: what goes whereKeep: A CLAUDE.md starterComing soon
  2. 5.2Hooks: lint gates, test runs, guardrailsKeep: Four hook scriptsComing soon
  3. 5.3Subagents: planner, implementer, reviewerKeep: Three subagent filesComing soon

Module 6 · An agent-ready repo · 3 lessons

  1. 6.1What makes a repo agent-readyKeep: A readiness checklistComing soon
  2. 6.2Scoring a repo for readinessKeep: A readiness scorecardComing soon
  3. 6.3Common gaps that block agentsKeep: The gap listComing soon

Module 7 · Security · 3 lessons

  1. 7.1Scanning code and dependencies for leaks and holesKeep: A security scan checklistComing soon
  2. 7.2Prompt injection and how agents get trickedKeep: The injection threat cardComing soon
  3. 7.3Using an agent to sort and fix security findingsKeep: A triage prompt and workflowComing soon

Module 8 · Agents at scale · 3 lessons

  1. 8.1Agents that work in the cloud while you do something elseKeep: When to send it to the cloudComing soon
  2. 8.2Running many agents at onceKeep: A fleet runbookComing soon
  3. 8.3From an issue to a pull requestKeep: A pipeline issue templateComing soon

Module 9 · Agents across a company · 3 lessons

  1. 9.1One door to your tools, with permissionsKeep: An access-policy templateComing soon
  2. 9.2Gateways, model routing and costKeep: A cost calculatorComing soon
  3. 9.3Rolling agents out across a companyKeep: An adoption playbookComing soon

Get the next lesson in your inbox when it's out.

Newsletter