Scenario-Driven + Test-First

Scenario-Driven + Test-First is the most critical principle in OpenLogos. It guarantees accuracy, reliability, and maintainability by combining two complementary strategies:

Scenario-Driven: Every feature is decomposed into discrete scenarios (S01, S02 …) that trace from requirements through design to implementation.
Test-First: For each scenario, a comprehensive test suite is designed before code is written, giving AI a precise definition of “done.”

The scenario journey

A single scenario ID (S01) travels through all three phases — no traceability matrix needed, because the ID is the trace:

Phase	What S01 gets
Phase 1 · WHY	Pain point, acceptance criteria (GIVEN/WHEN/THEN)
Phase 2 · WHAT	Interaction spec, page flow, prototype
Phase 3 · HOW	Sequence diagram → APIs → test cases (`UT-S01-001` …) → verified code

Change a requirement for S01? You know exactly which API, tests, and code are affected.

Test-first for AI

This is not traditional TDD (write one test → write code → refactor). OpenLogos designs the entire test system first:

Layer	Scope	Example
Unit tests (UT)	Function correctness	`UT-S01-001`: username min 2 chars
Scenario tests (ST)	Cross-module integration	`ST-S01-001`: full register → task view
Orchestration tests (OT)	End-to-end API flow	`OT-S01-001`: register → login → create task

Then AI generates code with a precise, verifiable target — “pass these 12 test cases with specific IDs,” not a vague “build a login feature.”

Why this matters

When you tell AI “write a registration feature,” it guesses what “done” means. When you tell AI “write code that passes UT-S01-001 through ST-S01-012,” it has a precise, verifiable target. The output quality difference is dramatic.

Automated verification

openlogos verify

Three-layer verification checks:

Check	Description
Pass rate	All tests pass
Design-time coverage	Every defined test case was executed
AC traceability	Requirements → test cases → results linked

The result is a Gate 3.5 acceptance report — quality as a number, not a feeling.

Test ID naming convention

{type}-S{scenario}-{number}

Type: UT (unit), ST (scenario), OT (orchestration)
Scenario: Two-digit scenario number (S01, S02 …)
Number: Sequential within the scenario (001, 002 …)

All test results are reported in JSONL format. See Test Results Format for the specification.