What this platform does not claim.

We keep the early stage honest. Every important result on this site links here. Maturity language describes HBF's handling and internal checks — not real-world validation.

FusionSim

• Cross-section source maturity. Registry entries carry validity domains and status; outside a source's domain, reactivity is flagged partial/invalid rather than silently used.
• Bremsstrahlung assumptions. Radiation models are approximations with declared validity in T_e; relativistic corrections are limited.
• Zero-dimensional model limits. L0–L1 are 0-D power-balance models; they omit transport, profiles, geometry and many loss channels. They are educational, not predictive.

DirectConvert — unsupported physics

• Space-charge and collective effects are unsupported at L0–L1 (warned, not modelled).
• The alpha-spectrum shape is an illustrative model, not an evaluated nuclear-data distribution.
• DC-L1 loss factors are assumed scalars with declared provenance, labelled unreviewed — never engineering predictions.
• DC-L2 trajectory studies are institutional-only and require reviewed field maps and independent code comparison (see the L2 validation matrix).
• No fabrication dimensions, insulation stacks, high-voltage operating procedures, vacuum-system construction, or component sourcing for dangerous systems.

AlphaTrack — synthetic-to-real gap

• All published metrics are on a synthetic dataset. A high synthetic score does not establish real CR-39 accuracy.
• Detector calibration is required before any species or energy inference: material/batch, etching protocol, microscope calibration, exposure geometry, filters, a partner-approved calibration curve, its validity range, an uncertainty model, and a named scientific reviewer.
• Out-of-domain or low-quality images trigger abstention ("manual review required"), which is a success, not a failure. Raw observed counts are always separated from calibrated estimates.

Evidence policy

• Marketing and public numbers are derived from released research objects — scientific numbers are never maintained separately in page HTML.
• Synthetic, unreviewed, extrapolated, stale and out-of-domain outputs are labelled with stable warning codes and cannot be silently exported as current.
• Source maturity (RAW → NORMALIZED → CURATOR_REVIEWED → VALIDATED → RETIRED) describes HBF's review of the record, not the correctness of every claim in a paper.

External review status

No external specialist review has been completed. All results are labelled unreviewed. Real-data validation, partner calibration, independent code comparison and institutional release remain pending (see the roadmap).