Both real-model answers satisfy the frozen controller, SSA identity, path-witness and machine-evidence oracle. Code Mode reduces cumulative input while increasing uncached input and runtime costs in this sample; preserve the complete mixed result and native-work confounds.
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Record correct real-model answers through complete atomic and corrected traditional native toolsets. Preserve the under-equipped initial baseline separately, full task costs, provenance, resource evidence and profile/strategy limits.
The direct model answers with machine facts alone; server Code Mode uses the new value queries but performs extra analysis. No inherent Code Mode or broad complex-software superiority is established.
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
Share exact coordinate selection across retained graph, projection and call-boundary values. Index inverse operand occurrences for values without requiring defining operations; preserve PHI slots, unknown spaces, partial scope and evidence.
Verify 250 OpenSSH SSA versions and 369 uses through both real MCP adapters. An equivalent retained-source task takes six core reads instead of 244; model costs and native speedups remain unproved. Pass 78 shared tests and actual catalog checks.
Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com>
Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com>
