{"schema_version":"v1","name_for_human":"AIDC-AI.IO Design Engine","name_for_model":"aidc_design_engine","description_for_human":"AI Data Center Infrastructure Design Engine — site selection, OPR/BOD, rackPlan, power, cooling, validation, cost, schedule, and procurement artifacts from one 3D graph.","description_for_model":"Use this tool when the user asks to size, validate, or lay out an AI data center anywhere in the world from a candidate site, building shell, IT load, rack profile, or existing 3D design graph. Three endpoints: - POST /api/agent/design — given IT load (MW), per-rack density (kW), GPU generation, site area (m²), region (metropolitan|regional), returns deployment-unit-snapped rack count, design PUE, PF-backed total MVA, Rubin liquid/air cooling split, CDU planning values, optional cost (KRW), and timeline (months). - POST /api/agent/validate — given a design summary or raw inputs, returns rule findings (electrical, cooling, layout, safety, data) classified blocking|warn|info, plus RFI items. - POST /api/agent/layout — given the design and optional site centroid, returns a rack-level plan (mm coords) and a site-level block plan (% coords). All preview calculations are deterministic graph projections. The engine derives OPR, BOD, standards matrix, rackPlan, electrical topology, cooling topology, building/site fit, validation/RFI rows, BOM, estimate, schedule, and procurement source data from the same 3D node/edge/schema graph. The AI agent analyzes the NVIDIA / OCP / Microsoft public AIDC facility manuals and reference set, recomposes those references with the selected site, rackPlan, power, cooling, network, AHJ/code, cost, schedule, and procurement constraints, then returns the strongest visual/system options. Electrical voltages, rack power interfaces, liquid/air heat split, CHW/TCS loops, CDU use, RDHx use, AHJ/code basis, vendor BOM, and procurement readiness are still selected and validated per project evidence; preview mode does not claim fixed universal equipment defaults. No key required for anon tier. Optional Authorization: Bearer aidc_live_<32 hex> upgrades tier when configured. Rate limit: /api/agent/* global limits: anon 10 req/hour, user 30 req/hour, registered 100 req/hour, partner 1000 req/hour; route burst guard: 20 req/min per IP.","auth":{"type":"none"},"api":{"type":"openapi","url":"https://aidc-ai.io/api/openapi.json"},"logo_url":"https://aidc-ai.io/icon","contact_email":"contact@aidc-ai.io","legal_info_url":"https://aidc-ai.io/terms"}