Benchmark Rerun Kits

These rerun kits are the shortest path from public benchmark package root to the strongest checked rerun lane for each workflow family.

The goal is not to list every artifact in the repository. The goal is to let an independent reviewer reopen one family without guessing:

• which package root is primary
• which companion package adds real pressure
• which runtime entrypoint actually counts
• which remaining limits still keep the family bounded

What Every Kit Must Provide

Each family kit should give a reviewer:

• one primary public package root
• one companion stress or transfer package
• one named runtime lane
• one clear run-mode statement
• one short list of remaining limits that still narrow the family sentence

Family Kits

family	current public posture	primary package	companion package	run mode	strongest remaining limiter
dda	outsider-auditable, bounded	dda_reviewable_run	dda_cross_engine_review_package	import_only	no in-repo live-engine parity yet
dia	outsider-auditable, bounded	dia_library_review_package	dia_matrix_shift_review_package	raw_executable	library incompleteness and downstream consequence still narrow the sentence
lfq	review-grade, bounded	lfq_cohort_review_package	lfq_sparse_contrast_review_package	raw_executable	missingness and cohort-transfer pressure still block stronger public language
multiplex	internal support only	multiplex_tmtpro_review_package	multiplex_channel_stress_review_package	raw_executable	outsider trust still collapses under the stress packet
ptm	outsider-auditable, bounded	ptm_localization_review_package	ptm_ambiguity_stress_review_package	raw_executable	consequence confidence remains weaker than localization strength
targeted	outsider-auditable, bounded	targeted_transition_review_package	targeted_carryover_review_package	raw_executable	calibration, interference, and burden still narrow stronger certainty

How To Open A Kit

For every family, use the same order:

1. open the primary package manifest and inventory
2. open the companion package manifest and inventory
3. open the named runtime rerun lane
4. inspect the rerun dossier or external review kit when the family publishes one
5. stop at the published limiter instead of improvising a stronger sentence

This is the point of the kit: a reviewer should not need maintainer folklore to know where the rerun route begins or where it still stops.

What The Kits Prove

• the current flagship families have explicit reopen paths rather than only descriptive trust prose
• paired-package pressure is part of the runtime story now, not a later documentation caveat
• run-mode honesty is visible at the rerun-kit layer instead of being buried in tests or implementation

What The Kits Still Do Not Prove

• scientific truth independent of the benchmark and grounding layers
• vendor parity just because a raw-executable lane exists
• downstream assay worth just because the rerun lane is real

Best Next Routes

• Open Operator Rerun Journey for the shortest operator-facing opening order.
• Open Execution for the family-level runtime ceiling.
• Open Benchmark Assets when the question becomes whether the public evidence root itself is strong enough.

Boundary

This page owns family-level rerun opening order. It should not turn into a second benchmark catalog or a second release-language page.