Spec-Driven Development
Book · Software Engineering × AI
Spec-Driven Development
Codifying AI agent guidelines. Eliminate the Ambiguity Tax in AI-assisted coding with machine-readable specifications, systematic plans and verifiable tasks.
By Muhammad Riaz
Start readingAI agent guidelinesTable of contents
Front Matter
Module 1: Foundations & the Ambiguity Tax
Chapter 1The Death of Ad-Hoc Prompting: Why AI Breaks at ScaleChapter 2The Ambiguity Tax: Measuring the True Cost of AI GuessworkChapter 3The SDD Maturity Spectrum: From Disposable Specs to Executable CodeChapter 4The Artifact Triad: Structuring `.spec.md` , `.plan.md` , and `.tasks.md`
Module 2: Authoring Machine-Readable Specs
Chapter 5The Anatomy of a Production-Grade `.spec.md`Chapter 6Setting Hard Boundaries: The Art of Explicit Scope ExclusionChapter 7Contracts, Schemas, and System InvariantsChapter 8The Interactive Clarification Loop: Using AI to Critique Your Specs
Module 3: From Spec to Plan — Systematic Execution
Chapter 9Architecting the `.plan.md` : Mapping File Impact and SystemsChapter 10Deconstructing Features into Atomic Checklists ( `.tasks.md` )Chapter 11Human-in-the-Loop Review Gates and Safety CheckpointsChapter 12Tooling Ecosystem: GitHub Spec Kit ( `specify` ), BMAD, and Agent CLI Workflows
Module 4: Agent Orchestration, Verification & Maintenance
Chapter 13Multi-Agent Execution Patterns: Implementors, Reviewers, and AdversariesChapter 14Automated Verification: Test-Driven Acceptance GatesChapter 15Preventing Spec Decay: Living Documentation and CI/CD PipelinesChapter 16Building an AI-Native Engineering Culture