Agentic EngineeringA field manual for building complex software with AI coding agents

Claude Code first. Senior engineers only. Mental models, playbooks, prompts, and failure catalogs — no hype, no beginner content.

Start Here

Prompt Library

Failure Catalog

Mental models, not metaphors

The agent as a compiler for intent. The repository as shared memory. Specs as the durable control layer. Models that change how you delegate, not slogans.

Foundations

Playbooks for real work

Step-by-step workflows for new projects, legacy refactors, production debugging, migrations, and tasks bigger than a context window — each with review points and recovery strategies.

Playbooks

A failure catalog that saves weeks

Seventeen documented ways agent projects go wrong — shallow demos, architecture erosion, mock-only tests, hallucinated APIs — with symptoms, early detection, and recovery prompts.

Failure Modes

50+ copy-paste prompts

Production-tested prompts for planning, design, implementation, review, security, recovery, and agent handoff. Each with required inputs, expected output, and evaluation criteria.

Prompt Library

Templates that do the governing

CLAUDE.md, AGENTS.md, ARCHITECTURE.md, feature specs, ADRs, PR checklists, handoff notes — the documents that keep agents inside your architecture when you are not watching.

Templates

Case studies with the mistakes left in

Ten realistic builds — MVPs, auth retrofits, CI/CD pipelines, multi-agent features — including the vague first prompt, what went wrong, and the intervention that fixed it.

Case Studies

Who this is for

Senior engineers, staff/principal engineers, CTOs, and AI-native builders who already know how to design, test, review, and ship production software — and want to do it through coding agents without losing engineering quality.

This site assumes fluency in architecture, Git, CI/CD, testing strategy, and code review. It never explains those. It explains what changes when a fast, tireless, literal-minded collaborator writes most of the code.

How to use this site

You need to…	Go to
Build the right instincts first	Foundations
Run a specific kind of task today	Playbooks
Write a prompt for a hard task	Prompt Library
Set up a repo agents can work in	Repository Design
Diagnose an agent gone wrong	Failure Catalog
Follow a structured curriculum	Learning Paths