[rohanarun]

This source-bound continuation raises conditional κ to 0.003644407895539, +0.000000140810139 over #7155 at 594d57036057f88f8bacc82e9c203bd4cb3276b9.

At chart 306, five timed gathers and seven later injections give exact helper vectors with c d = 1. Retirement at cut 36772 splits rank nine into four and five; one paid rank-eight path becomes ranks three and five. Two owned buffers and every added operation are charged. The native delta is T4 +1, T5 +1, T9 −1 and P3 +1, P5 +1, P8 −1.

The draft was published after source-bound native/operator and complete directed pricing passed. Full direct verification and registered isolated-cache replay now pass, matching every frozen receipt field with all 23 scientific hashes unchanged. Coverage includes whole forward/adjoint, the guarded temporal certificate with interior COPY-zero and final-owned obligations, 20 corruption controls, 5 sets of 512 chart/phase columns, finite physical replay, all 3,037 paid/helper maps and 104 main boundaries, ordinary/precision/fixed-tape costs, both directed prices, and 54 strict plus 4 functional outer conditions.

The unchanged 17-stage caller has 110,502 paid positions, 22,410 productive fields, surplus 1,424 and zero empty credit. Ordinary work is 781,880,696 < 10⁹; precision c = 2430 on grid 1125/F25. Owned labels/lifetimes/peak are 61/64/47. The existing 24-bit scalar contract passes. The independent directed crossing is strictly between 0.00364440789553997522914730314654796 and 0.00364440789553997522914730314654797.

python3.12 -B scripts/research_package.py --package research/p104-temporal-chart306-continuation

The complete 32-file #7155 export and all nested sources/notices are retained. Native canonical SHA256: b012295cfe44d204c2724fe544731ac17aa969e06e33043496ac7a1c91a912b3. PROOF.md, APPLICATION.md, VALIDATION.md, NATIVE-RESULT.json, RESULT.json and REPLAY.txt give the scope and evidence. Discovery stopped at its first positive, trial 4495 of the declared 12000-trial cap; no temporal optimum is claimed.

Credit Ali Romman for the causal-history identity; Jerry Chen/CubeJerry for retiming and verification; Matarener for kernels; dillona for the retained temporal constructor and source instances; Clear1nsight for the composition and guarded suffix certificate; ToikY for pricing; Rohan Arun for this instance, endpoint source, recharts and signed caller; and every retained author. Substantial OpenAI Codex assistance, no subagents; Apache-2.0.

Five inherited assumptions remain: native_all_columns, owned_erasure, compiler, precision_tape and analytic_recovery. Local verification does not establish hosted scientific CI, maintainer admission, independent human/formal acceptance or the complete multiplication theorem. The selected frontier is unchanged. The existing CI navigation correction from #7147 is retained and its targeted test passes; the separate historical Makefile-drift replay is outside this change.

Hosted scientific verification on the published head b44944c3058d522cbff17836e16240d40d77a927 passed: scientific replay job 130286697795. Exact native/price, full native, complete application and registered package gates all passed, with the registered PASS at 02:59:55 UTC on November 11. Overall CI remains red because the separate historical-selected-replay-gap job reports the previously disclosed working source drift: Makefile. Hosted scientific replay does not establish maintainer admission or discharge the five inherited assumptions.