liveBook · Manning

Specification-driven development

As explained in Vibe Engineering

A development approach in which an authoritative specification drives implementation plans, tasks, tests, and delivery controls.

A prompt can produce plausible code without establishing whether the result is safe to own. Specification-driven development addresses that risk by making the specification—not the prompt—the authoritative artifact from which plans, tasks, tests, and verification gates derive. section 1.3 section 1.4

The core rhythm is specify → implement → verify. An executable specification defines the intended effect before implementation and can span API schemas, error semantics, domain invariants, property checks, state-machine guards, performance gates, security policies, and provenance. A failed gate might be a broken contract, a violated invariant, a low mutation score, or excessive p95 latency—not just a failing unit test. section 1.5

A practical workflow separates exploration from production ownership:

  • Explore the domain through spikes, scoping, or prototypes before committing to a specification. section 1.3
  • Define intent and correctness before seeing generated implementation code; anticipate failure modes as well as the happy path. section 1.4
  • Derive and approve a plan and independently verifiable tasks. Vertical slices can cross policy, adapter, contract, and test, rather than adding one layer at a time. section 1.5
  • Have agents translate intent into code and verification, then enforce gates through CI/CD. Keep the human responsible for deciding which claims are true and auditing results. section 1.3 section 1.5
  • Iterate through verification and finish with tested, team-owned code—not merely a green build. section 1.3 section 1.5
Feature: Refund policy
  Rule: Time-bounded, single-order refunds
    Example: Within 30 days is allowed
      Given an order "O1" purchased at 2025-08-01T10:00:00Z
      When a refund of 25.00 USD is requested at 2025-08-28T09:59:00Z
      Then the refund is approved and remaining refundable equals 75.00 USD
 
    Example: After 30 days is rejected
      Given an order "O2" purchased at 2025-08-01T10:00:00Z
      When a refund is requested at 2025-08-31T10:00:01Z
      Then the request is rejected with reason "expired"
 
    Example: Partials must not exceed the original
      Given prior refunds of 90.00 USD on a 100.00 USD order
      When a refund of 20.00 USD is requested
      Then the request is rejected with reason "exceeds-remaining"

This executable refund specification makes approval, expiry, and cumulative refund limits concrete before implementation. section 1.5

The approach inherits the effect-before-implementation discipline of test-driven development, but puts a feature-wide specification at the center. Context engineering complements it by determining which instructions, code, documentation, constraints, and tool output reach the model. The model choice is secondary to the specification and its enforcement; generated code still requires verification and human ownership. section 1.3 section 1.5

How Vibe Engineering covers it

How deep it goes
Explains the specify–implement–verify rhythm, executable contract scope, human oversight, task slicing, and CI/CD gates.
Examples
Uses a refund policy with expiry and partial-refund limits to illustrate executable examples.
What you can do after
Practice a verification-first workflow that derives tasks and gates from a specification and audits agent-generated results.
How it connects
Connects test-driven development, context engineering, CI/CD, contract testing, and verification-first development.

Also explained in

Compare how 4 books explain specification-driven development →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Vibe Engineering. The section links point to the parts of the book it draws on.