← Back to T1ENS2 · Summon Your Hero — Prompt Engineering
T1ENS2 · Course material

S2 Lab · Live in Class (agenda)

S2 Lab · Live in Class

Done live — follow along, then repeat on your own agent afterward.

Four hands-on items. Bill runs each one on screen; you copy the moves into your own workspace.

① How to add an agent to your own website

What you'll do live: copy · paste · one embed line. Your agent goes live on your own site by the end of the session.

What you'll get:

  • The exact embed snippet (a single <script> tag).
  • Where to paste it (before </body>).
  • How to test it (open your site in an incognito window; the widget appears; type a message).

Why this matters: by the end of this Lab step, you have proof that ships. Not a slide, not a screenshot — a working agent on a URL you own.

② Heropedia and the one-click agent builder

What you'll do live: browse heropedia.org, find a hero someone else has already summoned, fork it into your own workspace, and ship it as your own agent.

What you'll get:

  • The Heropedia browse flow (search by domain, filter by hero style).
  • The fork mechanic (one click → your copy, editable).
  • The one-click "ship as my agent" button — from fork to live URL in under a minute.

Why this matters: you don't have to summon every hero from scratch. The community has done the hard work for the well-known ones. You stand on shoulders.

③ Multi-hero summon

What you'll do live: take one real question, and route it through multiple heroes in parallel. Each hero answers through their own lens. You get 360° coverage without having to write four separate prompts.

What you'll get:

  • The multi-hero pattern (one input → N Role prompts → N answers in one view).
  • A worked example using the three sages from today (Paul Graham · Peter Thiel · Marc Andreessen) on Jamie's case.
  • How to cross-strike — hand one sage's answer to the next sage as fresh input, and watch what surfaces.

Why this matters: this is the Sword-Duel pattern wired end-to-end. The pattern you saw in the slide deck, now runnable on your own questions.

④ Linus's CLAUDE.md explained

What you'll do live: open the actual Linus Torvalds CLAUDE.md — the industry-grade production prompt Bill runs every single day in his own coding workflow — and walk through it line by line.

What you'll learn:

  • Three iron laws — Good Taste · Pragmatism · Simplicity. What each one blocks. What each one enables.
  • Why it works — the three-layer dive baked into the prompt (Bug Phenomenon → Architecture Essence → Code Philosophy).
  • The "before vs. after" test — the same code question with and without this prompt. Night and day.
  • How to steal from it — which patterns to lift into your own hero's prompt today.

📎 The file we'll walk through: linus-torvalds-linux-md.md

Why this matters: every other Lab step teaches you how to summon. This step shows you what a summoning at the top of the craft actually looks like — so you know what to aim for.

After the Lab · what you take home

  1. A working agent on your own site (or on Heropedia).
  2. The multi-hero pattern, ready to paste into your own question.
  3. The Linus prompt, understood — not just copied.
  4. A cleared runway for the homework: everything you need to write your own hero for the domain you care about is now in your hands.

Slides won't do it for you — we build it live.