LIT is the Lispex interpreter used as a bounded dogfood and integration artifact. One canonical Lispex-in-Topaz source is exercised through the interpreter, generated Rust, generated Python, and Web routes to reveal real checker, lowering, runtime, and host-boundary defects.
What the evidence covers
The maintained integration records 84 of 205 primitive capability rows and six of 18 guest-calling rows. Unsupported rows fail explicitly. The four execution routes are variants of one source lineage, so their agreement is useful regression evidence but not independent implementations or proof of whole-language equivalence.
Product boundary
LIT is not installed by Topaz, is not a public backend, and does not add Lispex syntax, primitives, packages, or runtime promises to the Topaz language. Topaz releases may use the bounded court when a related compiler or host surface changes; full LIT campaigns remain release- or risk-triggered rather than an ordinary inner-loop cost.
How to read the result
A passing LIT court means the named source and admitted primitive rows behaved consistently on the recorded routes. It does not mean every Lispex program can run on Topaz, that the routes are source-independent, or that LIT is part of a Topaz installation. Unsupported capability rows remain the honest boundary.