Foundations - Start Here
Before you build the next thing, build the workbench. Five modules of basics every vibe coder needs: install the opinionated stack (VSCode, Claude Code, Node, Git, Docker), debug AI-generated code without re-prompting forever, and get real fluency with terminal, git, and environment basics. By the end, you have the same workbench the founder uses — and a debugging approach you actually trust. Around 3.5 hours of focused video. Built in public — content lands as it's filmed.
0%
The Full Stack
The AI builds you a full-stack app — frontend, backend, database — and you can't see the layers. This classroom fixes that. Fifteen lessons across five modules, one small todo app threaded through as the worked example. By the end, you can look at any AI-built app and name what you're seeing, and reason about where a problem actually lives.
0%
Hosting Ops
Hosting & Ops takes a vibe coder from "it works on a preview deploy" to "it runs on a real server I provisioned, at a real domain, for a cost I can predict — and I understand all of it."
0%
The Local AI Workbench
Stop renting your AI. Give your assistant a real map of your project, run a capable model you own with Ollama, and wire it into your dev loop — locally, for zero API tokens.
0%
Security Basics
Vibe coders ship fast, and security is the afterthought. AI-generated code makes it worse: it optimizes for "works," not "secure," and it won't flag the hole it just wrote. The classic disaster is mundane — a key committed to GitHub, scraped by a bot in minutes, the account drained before you've finished your coffee.
Nine lessons. The practical floor, not the academic ceiling.
0%
Unlock with Premium
Advanced AI Usage- The Action Chain
Premium opens here. Prompt-only AI fails in predictable ways: drift, unreliable structured output, hallucinated tool calls, no governance. The fix is structural. This classroom teaches the action-chain pattern, where code drives and the model is a typed gate that picks from a closed set, with every transition auditable. 32 lessons across 10 modules, each ending with a tested prompt you can run on your own codebase. You leave with a pattern that works on any product and any model.
0%
Unlock with Premium
Advanced Authentication
The auth layer above the basics. Sessions vs JWT, RBAC with the 404-not-403 pattern, multi-tenancy that fails closed, SSO, magic links, and safe account recovery. Six lessons, Hono and cookie-session shaped. Auth is where AI ships gaping holes, so every lesson ends with a verify checklist that is really a security review of the output.
0%
Unlock with Premium
Advanced Security
The engineering layer above Free's security hygiene. Seven lessons: STRIDE threat modeling, the OWASP Top 10 mapped to a Hono app, security headers with CSP and CORS, input validation against prompt injection, the 404-not-403 authorization deep dive, and a local OWASP ZAP rig in Docker Compose that gates your PRs. The cross-cutting capstone of Premium, where every verify checklist is the security baseline.
0%
Unlock with Premium
API design
Your API is the contract between your app and everything that touches it. Five lessons on the choices that age well: REST vs GraphQL vs RPC (REST is the default), URL versioning, deprecation with sunset headers, cursor pagination that holds at scale, and the minimum OpenAPI doc that pays for itself. Boring is durable. AI writes plausible APIs that miss versioning and deprecation, so every checklist names the part it skips.
0%
Unlock with Premium
Background Jobs
The day a slow request blocks a user, you reach for background jobs. Five lessons on doing it well at indie scale: when to queue and when not to, Postgres-backed queues with no new infrastructure, retries that demand idempotency instead of making failures worse, the four queue numbers worth watching, and running an action chain as a job. Queueing is a tax, so pay it late and with discipline.
0%
Unlock with Premium
Caching & Performance
The two-line rule of caching: add a cache only when measurement says you must, and remove it the moment that is no longer true. Most performance problems at indie scale are not cache-shaped. Five lessons: the three-axis test (is it slow, hot, stable), the right layer to cache at (rarely Redis), invalidation done honestly, the edge cases that quietly corrupt your cache, and the measurement that proves it earned its place. AI loves to suggest a cache. This teaches you when to say no.
0%
Unlock with Premium
Database Performance
Most app slowness at indie scale is database slowness, and most of that is a missing index or an N+1 query. Six lessons: reading a query plan, indexing that earns its place, killing the N+1 trap, SQLite to Postgres on the day the signals say so, the ten Postgres settings that matter, and partitioning only when measurement demands it. Try indexing first, then query restructuring, then caching. AI writes great schemas and indexes them wrong.
0%
Unlock with Premium
Front-End Engineering
Every other Premium classroom is backend-shaped, but vibe coders live in React. This is frontend engineering, not design. Seven lessons: component architecture, state at scale, the accessibility baseline AI keeps failing, performance on a real budget, a tiny design system, testing what matters, and streaming AI output to the UI. The gap between the component renders and it works for every user is wide and silent.
0%
Middleware Patterns
Middleware is where the request becomes context, and it's the substrate the rest of your backend runs on. Five lessons on the Hono stack: composability, one error envelope across every route, request context without prop-drilling, the lifecycle order that matters (logging, auth, rate limit, handler), and tests that catch order regressions. AI writes middleware that looks fine and fails silently, so every checklist targets those failures.
0%
Unlock with Premium
Production Engineering
The runway under your app: the operating discipline for a real SaaS at indie scale, 0 to 1,000 users, not enterprise SRE. 13 lessons across 5 modules: a CI pipeline you can read line by line, SAST and CVE scanning, the failure modes that actually happen and one-command rollback, a bill you can predict, observability sized for one person on call, and the honest signals that you have outgrown a single droplet. No Kubernetes, no SRE org chart.
0%
Webhooks + Integrations
Webhooks are async APIs you did not design: a request you did not make, from a sender you must verify, processed exactly once even if it arrives three times. Four lessons that make webhook handling correct on the first ship: signature verification with HMAC over the raw body, idempotency keys for exactly-once, the retry and replay defense, and the debugging discipline for the inevitable production issue. AI ships webhook handlers broken in subtle ways.
0%
The Free Vibe Companion
A drop-in companion folder for your AI coding assistant. Copy it into your repo, add one line to your CLAUDE.md, and your assistant starts applying Vibe Engineering's lessons, patterns, and recipes to your actual code. It walks you from empty folder to deployed SaaS: scaffold, database, auth, API, frontend, local AI, and a real server, writing docs from your real code as you go. You ship a product you understand and own, not a black box. Stack: React, Hono, Postgres, Drizzle, Ollama.
0%
1-17 of 17