REVIEW AND SCOPE
From Observable Answers to Executable Queries for CM/LM Programs
Technical report v1.1, 8 November 2026

The exact source received adversarial AI mathematical review. The final
revision explicitly requires nonnegative prior weights and qualifies the
3^m dynamic-programming count per initial observation fiber. The API guide
explains normalization within positive-mass fibers and use of an auxiliary
normalized prior to test feasibility in zero-mass fibers. The final scoped
review identified no outstanding mathematical issue.

A separate release assessment re-examined conflict covering, feasibility in
zero-prior branches, installed-union costs, the every-input addressing bound,
frame cancellation and the verifier's canonical-policy semantics. It found
no counterexample or mathematical repair to make within those assumptions.
This assessment does not establish acceptance by a human reviewer.

The public code checks finite examples; the general claims rest on the stated
proofs. The verifier certifies supported policy validity and truthful costs,
not optimality. The model assumes deterministic exact queries over finite
feasible states. No physical sensing, noise guarantee, learned embedding,
CM-exclusive speedup, comprehensive literature clearance or formal proof
certification is claimed. Human mathematical and editorial assessment remains
valuable. This is a technical/expository companion, not a journal acceptance.
