MarmicodeCharted Coding: AI-Assisted Development Without the Drift
1 Day · Online
🐣 Early Bird starts at €390
One day to move from fast-but-fragile AI coding to approaches you can sustain.Map the landscape, chart a method your architecture can live with — then wire harnesses that keep the agents on course.
🍱 What you will learn
Contrast Vibe Coding and Spec-Driven Development (Spec Kit, BMAD, OpenSpec, etc.) and tie them to a structured, test-driven approach.
Choose an AI-assisted workflow suited to prototypes, production features, greenfield, brownfield, or legacy code.
Write design documents that work for both humans and agents — without drowning in over-specification.
Align your intent with the code produced using tight iteration cycles.
Wire skills, hooks, tests, Nx boundaries, and ESLint rules as a deterministic verification layer the agent cannot quietly bypass.
Steer clear of drift, loss of control, over-engineering, review fatigue, and multitasking distraction.
Integrate these practices into daily workflows and clarify collaboration patterns around tests, prompting, and review.
Token spend is just the start. Cut the hidden costs: steering, reviewing, and rejecting AI output.
🗓️ Program
- Defining "Vibe Coding": when it works, why it is appealing, and classic pitfalls (drift, maintainability, illusion of productivity).
- Spec-Driven Development: Spec Kit (GitHub), BMAD, OpenSpec, and alternatives — anatomy and how Spec Kit works.
- Live demo: the same use case under both approaches — what holds up and what breaks.
- Comparative review as narrative: strengths and limits by context; when each pays off — and when it becomes a drag.
- Navigating with a map rather than drifting: steering the agent while keeping control of the trajectory.
- Chart the Intent: co-building a pragmatic Design Doc with the agent — goals, behavior, design, testing strategy.
- Plot the Waypoints: turning intent into an ordered, reviewable PR plan (thin slices that never break the mainline).
- Steer the Cycle: Scaffold → Red → Green → Refactor, with progressive review after each slice — not a big-bang review at the end.
- Tests as executable specification and as the AI agent’s feedback loop — how this differs from classic TDD.
- Compatibility with your current stack (Vitest, JUnit, pytest, etc.) — framework-agnostic mindset.
- Introduction to Charted Coding.
- Co-author a pragmatic Design Doc with the agent for the shared use case.
- Lock goals, non-goals, desired behavior, and high-level design before writing code.
- Turn the Design Doc into an ordered PR plan: thin, independently reviewable and mergeable slices.
- Identify scaffolding, pre-tidy-ups, and feature waypoints that keep the trajectory under control.
- Implement the first waypoint with Scaffold → Red → Green → Refactor.
- Progressive review: course-correct after each slice instead of drowning in a late mega-diff.
- Skills and hooks: packaging judgment so the agent follows your team's playbook.
- Verification gates: short feedback loops the agent must pass before moving on.
- Testing strategy as harness: executable specs (for real) that catch drift early.
- Nx module boundaries: architectural walls the agent cannot quietly cross.
- Set up or extend skills/hooks for the shared use case.
- Add or tighten a verification gate (tests, boundaries, or lint) and watch the agent respect it.
- Integrating the method into an existing team workflow.
- Collaboration patterns: who writes the tests, who drives the AI, who reviews.
- Choosing the right approach for the task at hand.
- Q&A and feedback from participants.
🎓 Required knowledge
- Development experience in a typed language (TypeScript, Java, C#, Python with types, etc.) — exercises use TypeScript
- Familiarity with automated testing
- Prior use of an AI assistant to generate code (Copilot, Cursor, Claude, etc.) — occasional use is sufficient
👨🏻🏫 Your instructor
Younes Jaaidi
I am a Software Cook who enjoys whipping code until tests pass.
With nearly 20 years of experience in eXtreme Programming, I've coached dozens of teams and trained thousands of developers to cook robust and maintainable software — using Test-Driven Development, pragmatic testing strategies, and a healthy dose of Collective Ownership.
I'm also an Angular Google Developer Expert, an NX Champion, and a mediocre sailor.
🙋 Frequently Asked Questions
Developers using or wanting to use AI assistants effectively; leads and tech leads framing AI usage; architects and CTOs industrializing AI-assisted development without sacrificing quality; and teams struggling with drift in generated code who want a structured, reproducible approach.
You should be comfortable in a typed language (exercises are in TypeScript), familiar with automated testing, and have tried an AI coding assistant at least occasionally.
A computer with internet access, microphone, webcam, an up-to-date browser, installation rights, and a working AI assistant (Copilot, Cursor, Claude Code, or equivalent).
Yes. After a lecture-and-demo comparison of the main approaches, you practice the Charted Coding workflow on a common use case — Chart the Intent, Plot the Waypoints, Steer the Cycle — then wire harnesses (skills, verification, boundaries). You leave with a collective synthesis and an individual action plan.
No. The principles apply across stacks; tests as executable specification are illustrated with patterns compatible with Vitest, JUnit, pytest, and similar runners.
Yes. Contact me for a quote and administrative details.
"Book a Session" lets you join a scheduled session with other participants. "Custom Session" is for companies who want a private, in-house workshop — with optional adjustments to content, duration, or focus areas.
If the workshop doesn't meet your expectations, reach out within 7 days and we'll work it out.