Skip to main content

Transcripts

Recorded walkthroughs of the labs. If your agent fails in the middle of a lab (sign-in, quota, network), open the lab's transcript and follow along: every prompt, what the agent did, and what came back. Then rejoin at the next module with your own agent.

Each was recorded during the rehearsal of the labs, with Claude Code, following the lab's instructions step by step. Results are shortened, and your agent's answers will differ in wording and sometimes in approach: that is normal. Where the lab has you edit a file by hand, the rehearsal had the agent do it, and the transcript says so.

LabWalkthrough
Lab 0 - Arrivallab-00-arrival.md
Lab 2 - Contextlab-02-context.md
Lab 3 - Skillslab-03-skills.md
Lab 4 - RobotCodelab-04-robotcode.md
Lab 5 - Prompt to greenlab-05-prompt-to-green.md
Lab 6 - MCPlab-06-mcp.md
Lab 7 - Hooks and the toolbeltlab-07-hooks-toolbelt.md
Lab 8 - Healinglab-08-healing.md, and the recorded healing report for the no-model path
Lab 9 - CIlab-09-ci.md, the workflows' output: Lab 9 needs no agent of your own
Bonus 1 - A librarybonus-1-library.md, self-paced after the day
Bonus 2 - A toolbonus-2-tool.md, self-paced after the day
Bonus 3 - Subagentsbonus-3-subagents.md, self-paced after the day
Cold opencold-open.md, the facilitators' backup for the first five minutes