Q Q Q Q Quality A A A A Assurance
Traceability, conformance, and knowledge governance
Quality is typically an afterthought — manual code review, ad-hoc testing, and "it looks right" as the acceptance bar. Teams rely on individual discipline rather than systematic enforcement, and the result is predictable: bugs that should have been caught in design surface in production, patterns that were solved months ago get reinvented, and the codebase accumulates technical debt that no one can trace back to a decision.
SBX builds quality into every step of the pipeline. 352+ knowledge artifacts are loaded automatically at the relevant SDLC step. Conformance gates verify traceability from story to test to code — and they block, not warn. The solution-before-code rule ensures proven approaches are reused, not reinvented. Quality is not a phase; it is the structure itself.
352+ Governed Knowledge Artifacts
The knowledge system organizes institutional memory into five artifact types: patterns (reusable design approaches), standards (mandatory coding and modeling rules), guidelines (contextual best practices and proven implementations), toolkit (platform-specific tools and libraries), and NFRs (non-functional requirements applied per-entity). Each artifact is loaded contextually at the relevant SDLC step — developers do not need to search for best practices or remember which standard applies. The pipeline delivers the right knowledge at the right time, every time.
From Story to Deployment — Nothing Lost
Full traceability means every exported function traces back through its test, fixture, scenario, and user story. The sbx trace command verifies this chain mechanically — no manual audits, no spreadsheets, no "I think that function is for the login feature."
Gaps are detected at FDD5.CONFORMANCE and FDD5.TRACE. Code without a story has no reason to exist and can be deleted. Code without a test has no proof that it works. This chain is not documentation — it is the structural integrity of the codebase.
User Story
The business requirement and acceptance criteria that justify the work. Every feature starts here — if there is no story, there is no reason to write code.
Scenario
Given/When/Then behavior specifications that decompose the story into testable interactions. Each scenario is a concrete contract between intent and implementation.
Test Fixture
Concrete input/output data pairs derived from scenarios. Fixtures are the bridge between human-readable requirements and machine-verifiable assertions.
Unit Test
Automated verification of behavior using fixture data. Tests are written before the function exists — red/green/refactor, no exceptions.
Function
The implementation that satisfies the test. Every exported function traces back through this chain. If the trace breaks, the function has no reason to exist.
Package
The module boundary containing the function. Packages enforce separation of concerns and define the public API surface.
Application
The deployed artifact that composes packages into a running system. The final link in the chain — from requirement to production.
Quality Gates That Block, Not Warn
Conformance is mandatory, not advisory. At FDD5.CONFORMANCE, the system verifies model-code conformance — every exported function must trace to a user story. At FDD5.TRACE, traceability is audited across the full chain. These gates block completion: you cannot ship without passing them. This is not bureaucracy; it is insurance against the "quick fix" that breaks everything downstream. A warning is a suggestion. A gate is a guarantee.
Build Verification
Compile passes with zero errors, zero warnings. Both Go and Svelte platforms are checked. If the build fails, nothing else runs — this is the first gate and the most basic.
Conformance Gate
Every exported function traces to a user story. Model-code conformance is verified — the implementation must be a faithful representation of the specification, not an approximation.
Traceability Audit
No gaps in the story-to-test-to-code chain. The sbx trace command walks every link. Orphaned code — functions without stories, tests without scenarios — is flagged for deletion.
Regression Walk
Visual verification via Playwright screenshots. Layout requirements are measured against actual rendered output. This catches the bugs that unit tests cannot — the ones you can only see.
Quality Commands
Knowledge loading, traceability verification, and regression testing are all driven from the terminal. Select a command below to see its usage and output.