One inspectable design source becomes OpenVSP geometry, OpenAeroStruct aerostructures, OpenMDAO optimization, and an evidence-rich design review you can challenge down to the exported mesh.
Featured · AtomRC Dolphin V1.1Optimized top + side · open the interactive review →
Each review binds a baseline aircraft and its optimized descendant to phase-local geometry, aerodynamic, structural, balance, and provenance evidence. Open a card to orbit the delivered mesh and inspect every gate.
Scan-grounded reproduction with measured mass and CG, an eleven-section wing loft, buried fin roots, and travel-bounded elevon pitch trim.
Provisional parametric sport-jet reconstruction traced from an orthographic family drawing and photo cross-checks.
Source-locked reconstruction of NASA's 5.5-percent AirSTAR Generic Transport Model with equivalent twin-engine propulsion.
Source-only flying-wing reconstruction with recorded flight-dynamics evidence and phase-bound simulator delivery artifacts.
One-third-scale sailplane reconstruction with flexible-response evidence and an explicit record of its validation boundary.
Full-scale transport reconstruction publishing geometry, aerodynamics, and propulsion while preserving the recorded wingbox-closure allowance.
Sealed 2026-08-22 NTNU X8 holdout artifact retained as a frozen report; it has no mutable design source or regenerable package in this revision.
Python is the control plane; established geometry, aerodynamics, structures, optimization, and stretch-solver kernels stay intact. The source YAML remains authoritative throughout.
Requirements, sketches, or a reference mesh become one schema-checked design.yaml.
Sizing, OpenVSP geometry, aerodynamics, flight dynamics, and structures evaluate the baseline aircraft.
OpenMDAO searches bounded design variables, then the delivered aircraft is independently re-evaluated.
Twelve gates and optional TACS, VSPAERO, and SU2 cross-checks expose assumptions, failures, and model limits.
A green pipeline is the beginning of review, not the end. Every headline value must trace to the same aircraft phase and every validation statement stays inside its recorded evidence class and intended use.
ok: true is a stage claim, not a verdict. Baseline and optimized results describe different aircraft and are never mixed in one claim. TACS and SU2 remain calibration cross-checks, not pass/fail evidence.