AI Agent Engineering Practice

How to build, validate, maintain and evolve complex systems with coding agents, without confusing passing tests, completed documents and delivered outcomes.

The on-demand engineering methods connect design review, staged delivery, evidence, diagnosis and retirement. The staged weekly-review case study, retiring obsolete engineering knowledge, and refactoring, retiring, or deleting tests when writing a Spec provide concrete workflow lessons. See the roadmap for genuinely unfinished topics.

CURATED READING PATHS3 categories · 63 articles

AI-ASSISTED ENGINEERING

01 / AI Agent Engineering Practice

How coding agents develop, validate, maintain and evolve software.

8 articles →

01 / 02

Specs, Evidence and Engineering Memory

From specifications to execution evidence and durable project history.

1. Specs, Evidence and Verifiable Engineering Closure ↗
2. Why We Split a Large Weekly Review Spec into Three Phases ↗
3. Engineering Knowledge: Context, Memory and Retirement ↗

02 / 02

Real Validation and Task Convergence

From misleading green tests to verifiable task completion.

1. Start with a Report You Can Read Back: Open ETF Agent v0.1 ↗
2. Tests and Evaluations: What Does Green Prove? ↗
3. Tests Expire Too: Move Refactoring, Retirement, and Deletion into the Spec ↗
4. Why Did One Subagent Change Hold Up Our Work for an Entire Day? ↗
5. From Bug-Fix Loops to Task Convergence ↗