Evidence before interpretation
How readings are produced
Kundli Circle separates deterministic Engine calculations, model-assisted interpretation, and human expert activity so their provenance remains clear.
Engine facts
Canonical birth snapshots are sent server to server to Kundli Engine 0.14.0 under contract 2.11.0. The application stores calculation IDs, input and response hashes, ruleset versions, warnings, availability, and immutable response snapshots.
Interpretation
A server-only structured interpretation provider receives minimized Engine facts and explicit customer context. Every material claim must cite permitted evidence IDs. Unknown evidence, prohibited certainty, fear language, or schema failure blocks delivery.
Expert Reviewed
An assigned astrologer sees read-only Engine facts and the interpretation draft. Approval creates a new immutable customer-visible report version. A suspected calculation issue must be escalated; it cannot be silently overwritten.
Birth-time uncertainty
Exact, approximate, and unknown birth times can make different calculations available. Reports preserve these limits and never upgrade confidence beyond the Engine response.