StackBuilder deterministically composes 4+1 reference architectures from current Layer2C assessment grades. It never re-grades evidence: capability and authority are quoted from the canon, never generated. Given identical inputs, it produces identical architectures.
The application layer of the Layer2C research system. It consumes assessed evidence, it does not create it: the 4+1 vendor-authority grades from layer2c.com and Fourth Cloud readiness from cloud.layer2c.com. It composes vendors layer by layer; the model narrates the result, it never determines it. It is an educational instrument that names the trade, not a certification or a vendor ranking.
Every composed layer cites its source assessment on layer2c.com. The reference reports extend that to the hands-on Lab (labs.layer2c.com) that validated the vendor at that layer where one exists. Grades are never modified here.
Identical inputs including an explicit anchor produce a byte-identical composition; without an anchor, tie-breaks among equally-graded leaders are randomized, so pass the anchor from an output back in to reproduce it. The exact pick algorithm, affinity-lock, and Design Risk and RDI formulas are published at /stackbuilder-reproducibility.json, so another agent can reconstruct a stack and verify every signal. Saved roadmaps (account required) store the inputs plus the resolved anchor and re-derive from current grades, so an assessment update flows through; they are reproducible by input, not frozen by version.
The graded 4+1 vendor assessments at layer2c.com and the Fourth Cloud readiness assessments at cloud.layer2c.com. The available components (vendors and layers) are listed at /components.json.
Request: {"vendors":["aws"],"objective":"capability"}. Result: AWS covers all eight layers, coverage 8/8, zero integration seams, Design Risk low, but all eight layers are Ceded, so no authority is retained, and RDI is high (the stack acts on a moderate reasoning plane it does not own). The full report, with the assessment-and-lab evidence chain per layer, is at /example-composition.json.
StackBuilder composes reference architectures from Layer2C assessed evidence. It does not ask an LLM what vendors can do. The assessments are the source of truth; the model explains the recommendation.
The engine is deterministic. It reads the graded assessments from Layer2C (vendor authority across the 4+1 Layer AI Infrastructure Model, scored with the Decision Authority Placement Model, DAPM) and Fourth Cloud (on-prem control-plane readiness, FC-0 through FC-4), composes vendors per layer, and computes two signals: Design Risk (integration effort, where the seams sit) and the Reasoning Dependency Index (RDI), which reads how much autonomy the composed stack exercises. It never re-grades an assessment.
Recommendations are bounded by available assessments and methodology. Missing evidence should stay visible rather than being filled in by model confidence.
It's an educational instrument. It names the trade: the control you cede, the gaps you own, the integration effort you carry. It sharpens your conversation with an architect. It isn't a turnkey design.
The application is client-rendered, so its inputs, rules, and outputs are also published as machine-readable files a crawler can read without running the app:
Layer2C is an independent architectural research system. Assessments are opinionated analyses, not certifications, paid rankings, procurement recommendations, or permanent conclusions. Machine-readable index: llms.txt. This page is an interactive application; enable JavaScript for the full instrument.