Vibe coding to production

For people who build with coding agents and app builders. Keep the human gates that make agent-built software safe to ship: a brief, a reviewed diff, tests that prove something, least-privilege tokens and a rollback. Already built something? Module 2 starts from your app.

Modules in this path

1. 1. How coding agents workThe loop, and why agents are confidently wrong
   1. How coding agents actually work Reading, 5 min
   2. What agents are good and bad at Reading, 5 min
   3. Check: how agents work Knowledge check, 5 min
   3 open
2. 2. Already built it? Start hereFrom an app builder to a repository you run
   1. Already built it? Start here Reading, 6 min
   2. Run it on your machine Reading, 6 min
   3. Why it works locally and not online Reading, 5 min
   4. Check: from builder to repository Knowledge check, 5 min
   4 open
3. 3. Brief before promptScope, acceptance criteria, small steps
   1. Brief before prompt Reading, 6 min
   2. Small steps, one change at a time Reading, 6 min
   3. Check: briefs Knowledge check, 5 min
   3 open
4. 4. Read every diffReview an agent's pull request
   1. Read every diff Reading, 6 min
   2. Review an agent's pull request Reading, 8 min
   3. Check: reviewing diffs Knowledge check, 5 min
   3 open
5. 5. Tests that prove somethingTests that fail first, then pass
   1. Tests that prove something Reading, 6 min
   2. Let the agent write tests, then check them Reading, 6 min
   3. Check: tests Knowledge check, 5 min
   3 open
6. 6. Security of AI-written codeSecrets, authorization, injection, invented packages
   1. Secrets don't belong in code Reading, 6 min
   2. Authorization: who may do this? Reading, 6 min
   3. Injection, invented packages and licences Reading, 6 min
   4. Check: security of AI-written code Knowledge check, 5 min
   4 open
7. 7. Give your agent contextInstruction files, skills and MCP
   1. Give your agent context Reading, 6 min
   2. Skills and MCP Reading, 6 min
   3. Check: context Knowledge check, 5 min
   3 open
8. 8. Ship it with an agentA scoped, expiring token and a reviewed deploy
   1. Least-privilege tokens Reading, 6 min
   2. Deploy the starter with an agent Reading, 6 min
   3. Lab: ship the starter Verified lab, 35 min
   4. Check: shipping Knowledge check, 5 min
   4 open
9. 9. Operate, cost and when to stopLogs, rollback and a human in the loop
   1. Operate with an agent, safely Reading, 6 min
   2. Cost and when to stop Reading, 5 min
   3. Lab: roll back an agent's bad release Verified lab, 30 min
   4. Check: operate Knowledge check, 5 min
   4 open