[aliromman]

This enabling proof exposes 222 original helper queries before the retained forward BIT source starts, with two queries at each of 111 ports. It proves the actual 111 port routes and preserves the source endpoint and native paid profile. The complete kappa calculation does not improve: a profitable caller, its coupled-helper decoding, and its ordinary-work tariff remain open.

The proof, exact result, and source pins state the boundary. The publication base is 32890bc5de034ab8fc93ac88ead7279a6a416d04. The four live inputs are unchanged from the original investigation base 1ec1119972c9aefe42525172d07b332fcff81cb4.

The direct interface checks all 661 persistent input columns and 445 owned input columns through all 222 replacement cuts. Its additional work is 444 owned COPYs, 888 owned CLEARs, 222 whole-field SWAPs, and at most 444 owned fields. A fresh global stage also uses 666 fixed cover-route maps. No native child is added. A separate optional 286-query local variant has 572 COPYs, 1,144 CLEARs, 286 SWAPs, 572 owned fields, and 843 head chart maps; those are different counts and contracts.

The retained word comes from PR #589 and PR #555. The source notice credits eumemic with substantial Anthropic Claude assistance and lists the earlier contributors. The entrance prologue follows PR #5388 and its pinned proof; that proof is conceptual credit, not an executable dependency. PR #5421 instead uses nine owned relations in the reverse word. This package keeps arbitrary forward helper inputs and their target debits. The BIT adaptation, proof and checker have substantial OpenAI Codex assistance; Apache-2.0.

Run with Python 3.12 and no third-party packages:

python3 -B research/bit-early-helper-queries/verify.py
python3 -O -B research/bit-early-helper-queries/verify.py
python3 -B scripts/research_package.py --package research/bit-early-helper-queries --check-only
python3 -B scripts/research_package.py --package research/bit-early-helper-queries

Expected direct output:

PASS_BIT_EARLY_HELPER_QUERIES queries=222 routes=111 source_controls=4 owned_controls=5 new_kappa=false

Normal and optimized checks passed with identical reports. The Python 3.12 registered replay, committed-byte check, committed export check, and normal/optimized replay in a fresh Git-free export passed. Controls reject all four source-byte mutations, a wrong held input, a wrong frame-tag swap, a dirty owned input, an uncleared raw field, and an uncleared observation.

The optional source-history preflight could not pass because this checkout is shallow; the initial scientific commit directly follows the declared base, and all live inputs were included in the successful Git-free export. Hosted CI has not run at publication. No native payload, full GL cover, uniform compiler, ordinary precision, or outer multiplication assembly was run. The open next questions require a complete caller before further cost claims.