[dillona]

Rohan Arun's #7122 rechart makes helper 1445 share an input boundary with helpers 454, 1208 and 1461. The resulting basis permits an additional rank-eight helper split into 7+1. This raises conditional κ to 0.003641239650184, a gain of 0.000000478276042 over #7122.

At exact #7122 cut 40552/chart 924/pivot 454, one rank-one MOVE is neutrally advanced. The split adds 22 ADDs and four SCALEs, with no paid-profile increase. The complete signed 17-stage application has 110,655 paid positions, 22,291 productive fields, surplus 1,424 and no empty credit. Ordinary source-unit work is 777,763,024 and fresh precision remains c=2344. Native 1125/F24 and caller 675 conversions, physical ownership and fixed-tape costs are all included.

Both complete direct CPython 3.14.7 commands and the final registered isolated CPython 3.12.16 replay passed. Every frozen native and composition receipt field matched. REPLAY.txt preserves the actual output; metadata was marked completed afterwards.

Coverage includes exact source/candidate reconstruction, arbitrary-input forward/adjoint, 17 native corruption controls, four 512-column chart checks, finite native/signed routes, rational/integer Laurent caller, all 3,048 paid/helper maps and 104 main boundaries, fresh envelopes and complete cost substitutions, both directed price engines, a crossing-probe rejection and 54 strict plus four functional outer conditions. The two integration stops and their resolutions are recorded in VALIDATION.md.

The full nine-file #7122 export at 98aad9eb0e7d7f985622a35c7981695fe680c68a is frozen with its entire nested #7121 closure. The author's later completed revision 6391d64cd0e37e0988e6b9c41fe13f053ed2c81e retains identical construction/source files; PEER-STATUS.json distinguishes that status from this package's own replay. Both engines reject #7122's invoice at the new claim. The independent raw crossing is (0.00364123965018407336305001489908943, 0.00364123965018407336305001489908944)—a bound for this emitted invoice, not other constructions.

python3.12 -B scripts/research_package.py --package research/p104-recharted-rank8-split

The five inherited assumptions, independent human/formal review, maintainer admission and compiled top-level execution remain. No selected-frontier change. Hosted CI is separate from the local evidence; the known canonical guide6030/selected7000 onboarding mismatch remains outside this package-only change.

Credit Rohan Arun for #7122's rechart and #7113's caller/application; 1337isnot1337/OpenAI Codex for the fresh-helper-boundary alpha-expansion method; Ali Romman, Jerry Chen/CubeJerry, Matarener, ToikY, DreamingOfClouds, Douglas Colkitt/#7000 reviewers and every retained source notice. Substantial OpenAI Codex assistance; no subagents. Canonical investigation.

Exact-head hosted run42903313790 failed onboarding at21:19UTC (760tests,1failure,24skips): guide6030 versus selected7000 in test_current_ci_section_matches_selected_record. Scientific replay was skipped. This is separate from the successful local registered evidence.