[JoyboyBrian]

The actual Raw.shuttle owner uses numeric LEFT/BLANK framing, while the completed binary elementary interchange expects aligned three-bit ports. This package supplies a literal paid encode/init/run/install/decode/retire boundary that returns the exchanged payload to the same RAW owner and retires every encoded bank. The source program and frozen predecessors remain unchanged.

The source-relative proof covers admitted widths H >= 0, A,G,S >= 1, with complete local instruction and per-role retained-backing bounds. The 29 roles include the RAW owner, both actual descriptor banks and 20 encoded child roles. The child uses {raw BLANK4, ONE1}; the enclosing caller remains five-symbol. Actual standalone u/t backing is 3,146,057 cells each; the shared u→t per-role invoice is 3,146,075, demonstrating the 18-cell undercount from taking only the maximum standalone total. The t/history figure is theorem arithmetic, not an executed source-sized history.

Validation on CPython 3.12.2/macOS:

• Fresh normal and optimized light replays pass 24 small/history/basis calls, two literal interpreters, exact retained capacities, wrong-decoded-bit and omitted-install controls. Optimized Git-free missing/changed-source and typed-receipt controls reject.
• Package metadata, committed bytes, executable modes and source-history ancestry pass. The official export covers all eight live inputs and 21 content files; both normal and optimized Git-free read-only replays pass with all 22 exported files, including the manifest, unchanged.
• The preserved historical FULL.json records the actual 196,608-bit u call: 336,273,602 shuttle instructions and 3,146,057 retained cells, plus 281 descriptor initialization and 389 retirement instructions. Its bytes are unchanged. No local full-u replay was run during this publication refresh. The registered hook includes a separately budgeted 180-second --full command; hosted CI is separate evidence.

Reproduce the bounded checks:

python3.12 -B scripts/research_package.py --package research/raw-crt-binary-shuttle-bridge --check-only
python3.12 -I -B research/raw-crt-binary-shuttle-bridge/verify.py
python3.12 -I -O -B research/raw-crt-binary-shuttle-bridge/verify.py
python3.12 -I -O -B research/raw-crt-binary-shuttle-bridge/test_integrity.py

This is one replacement call boundary, not an execution of the full original Raw class. Source-sized t, the complete twelve-shuttle/rotation/ROM/repair route, upstream payload production, whole-caller binary lowering, global allocation, native supplier, precision and outer multiplication remain open. No malformed-input rollback, recursive invoice change, new κ, complete theorem, human acceptance or formal-kernel verification is claimed. The prior scoped AI review is historical. A fresh separate AI source/proof review at this publication head checked symbol renaming, same-owner installation, all29-role lifecycle, per-role maxima and the instruction tariff; an additional dual-interpreter growing/shrinking history passed111,426 instructions. Hosted CI remains separate evidence.

The two original research commits are preserved by cherry-pick onto canonical main 721c03822c421b7f7fcbb2598993b947c4d88419; a third documentation commit records the resumed claim and historical evidence boundary. All scientific executable/proof/receipt bytes and eight live input hashes match the original result. Original revision-7 provenance stays pinned; the current checkpoint is priority revision 8 with selected PR6030. Overlap checked 341 open PR bodies, 5,722 Discussion records and 8,348 tracked local notes; no exact successor was found. This snapshot does not cover every closed PR or unpublished result.

Canonical claim. Credit: Jose Manuel and retained PR2809/2753 authors; JoyboyBrian/PR5162/5188; Ali Romman for binary lowering; Hiroki KUMAZAKI/PR5286 for retained backing; Cinna Davis, Douglas Colkitt/OpenAI and the retained CRT lineage. New bridge and publication by JoyboyBrian with substantial OpenAI Codex assistance, Apache-2.0.

[JoyboyBrian]

CI follow-up for 909f5c08babc24622a4b344e32b50274317ef011.

The plan failure reproduces on clean canonical main 721c03822c421b7f7fcbb2598993b947c4d88419: the two current-status navigation tests fail before this PR’s scientific replay. Existing PR #6050 (41ac3a47b91b4184e6c54af417bf576bfb429926) fixes both documentation mismatches; its exact two-file change passes all six tests locally, normally and with -O. Awaiting the shared upstream repair and a fresh CI run; this is not a successful scientific-replay claim.