Join a Claude Community Event, organized by local enthusiasts for everyone who loves building with Claude.
Build with AI. Own the Evidence.A hands-on Claude Code workshop: build, break, and fix your own hackathon idea coach
Description
Most people use AI on autopilot: ask, accept, ship. This workshop runs the other way. Working in pairs, you'll use Claude Code to build a small AI coach that pressure-tests a hackathon idea. It asks who actually has the problem, counts the people you can reach, refuses to invent research, and helps you design one small experiment before you commit to building. You write the spec. Claude implements it. Then you try to break it and repair what fails. You leave with a coach you built yourself, test evidence you can explain to a judge, and a spec-then-verify workflow you can reuse on your hackathon project that same weekend. Bring a laptop. Claude Code access is provided through Anthropic API credits. No prior AI or coding experience needed. CS and business students both welcome.
Agenda (90 minutes)
0:00 Copilot vs autopilot Why owning the decision matters, and what you'll build today
0:06 Meet the reference coach See the target behavior and its four outputs: idea brief, count ledger, experiment card, MVP brief
0:12 Checkpoint 1: set upCreate your build folder, start Claude Code, have it read the brief
0:20 Checkpoint 2: spec before codeDraft SPEC.md, challenge one rule with your partner, approve it
0:32 Checkpoint 3: buildClaude implements your spec: prompt, four templates, test cases
0:48 Checkpoint 4: test and repairRun adversarial inputs, judge the answers, fix one failure, retest
1:12 Engineering lensesDRY, SOLID, ACID, Big O, and where BMAD and gstack fit1:22Show your workEach pair demos their coach and takes home the build kit
