WiseSeatAI Guide
Spec-Driven Development: The SDD Process for AI-Assisted Teams
Spec-Driven Development (SDD) treats the specification as the source of truth for AI-assisted delivery. Requirements, constraints, acceptance criteria and engineering rules are written before implementation so coding agents work inside a controlled system.
Spec-Driven Development lifecycle
- 01Constitution
- 02Requirements
- 03Specification
- 04Technical plan
- 05Tasks
- 06Agent implementation
- 07Tests and verification
- 08Human approval
- 09Specification update
SDD versus vibe coding
Vibe coding starts with a prompt and accepts whatever the agent can produce. Spec-Driven Development starts with intent, rules and acceptance criteria, then asks the agent to implement against that written contract.
The difference matters for teams because shared specifications make work reviewable, repeatable and easier to hand between developers and agents.
SDD versus TDD and BDD
Test-Driven Development focuses on tests as the design driver. Behavior-Driven Development focuses on behavior examples and shared business language. Spec-Driven Development can include both, but uses a broader specification as the operating context for planning, implementation and review.
Core artefacts
An SDD workflow usually needs a repository constitution, a feature specification, a technical plan, implementation tasks and validation evidence. These artefacts tell agents what matters before they edit code.
- Constitution: stable engineering rules and architecture constraints.
- Specification: user intent, scenarios, constraints and acceptance criteria.
- Plan: technical approach, affected surfaces and verification strategy.
- Tasks: small implementation units with clear completion checks.
Example for a Java/Spring service
For a Java/Spring service, the specification should define API behavior, domain rules, persistence expectations, error cases, security requirements and migration constraints before an agent starts implementation.
The agent can then work through controller, service, repository, migration and test changes while the reviewer checks every output against the specification and project constitution.
SDD readiness checklist
- Create repository-level engineering rules.
- Write acceptance criteria before implementation.
- Keep agent tasks small enough to review.
- Require automated verification evidence.
- Update the specification when code behavior changes.
- Make human approval explicit before delivery.
Frequently asked questions
What is Spec-Driven Development?
Spec-Driven Development is a software delivery process where the written specification guides planning, agent implementation, testing and human review.
Is SDD the same as TDD?
No. TDD centers development around tests. SDD uses a broader specification that can include tests, behavior examples, architecture rules and acceptance criteria.
How does SDD work with AI coding agents?
The specification becomes the agent context. It tells the agent what to build, what constraints to respect and how humans will verify the result.