Workflow Claim Limits¶

This page is the release-facing claim-limit source of truth for the flagship workflow families.

Outsider-auditable workflow families today: dda, dia, ptm, targeted. Released outsider-auditable workflow families today: dda, dia, ptm, targeted. Internal-support-only workflow families today: multiplex. Full outsider-readable family packets today: dda, dia, lfq, ptm, targeted.

Family-level trust now requires two public benchmark packages plus one published cross-package generalization report. The current scorecard lives at packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/family_stability_scorecard.json.

What changed since v0.3.7 is not only wording. The matrix now sits on a deeper chain of proof: broader core scientific surfaces, clearer runtime rerun lanes, explicit knowledge grounding, recommendation posture, and visible lab consequence pressure. That means the repository can now distinguish between families that have a full outsider-readable packet and families whose released sentence must still stay narrower.

The point of this matrix is to keep that stronger depth from being flattened into one generic trust label. A family can now have meaningful benchmark, runtime, grounding, recommendation, and consequence surfaces while still stopping short of the broadest released sentence.

What This Matrix Governs¶

what the repository can publicly say today for each workflow family
which family has a full benchmark, runtime, grounding, recommendation, and consequence packet
which family still narrows at the released sentence because one part of that packet remains downgrade-heavy
which family stops at internal support even though code, runtime, and public benchmark substance exist

The distinction that matters most now is this:

outsider-auditable is the stronger release-language set
full outsider-readable family packets is the wider proof-packet set
LFQ currently belongs to the second set but not the first

Why This Distinction Matters¶

it keeps the repository from flattening all strong-looking family packets into one trust label
it explains why a public packet can be openable and still remain narrowed at release time
it makes clear that consequence and burden surfaces can still veto broader family language

Current Limits¶

Workflow	Internal benchmark-backed	Raw-executable	Externally cross-checked	Outsider-auditable	Lab-consequential	Public language
`dda`	yes	no	yes	yes	yes	bounded outsider-auditable
`dia`	yes	yes	yes	yes	yes	bounded outsider-auditable
`lfq`	yes	yes	yes	no	yes	review-grade bounded
`multiplex`	yes	yes	no	no	no	internal support only
`ptm`	yes	yes	yes	yes	yes	bounded outsider-auditable
`targeted`	yes	yes	yes	yes	yes	bounded outsider-auditable

How To Read This¶

internal benchmark-backed means a tracked flagship public package exists under benchmark-assets/flagship-public-packages/
raw-executable means the strongest current runtime lane runs the tracked package directly instead of stopping at an import-only bridge
externally cross-checked means public comparator posture is at least advisory rather than refused
outsider-auditable means the benchmark package, runtime lane, scientific reading, recommendation packet, planned lab packet, requested-versus-observed outcome dossier, and the family generalization report can be opened together by a skeptical reviewer
review-grade bounded means the benchmark, runtime, and review surfaces stay usable, but the current external review kit or acceptance pressure still narrows the stronger outsider-facing sentence
lab-consequential means a dedicated flagship lab packet, one shipped requested-versus-observed outcome dossier, and one assay-worth-it ledger row all exist together

What Outsider-Auditable Still Refuses¶

it does not mean universal transfer across cohorts, instruments, study designs, or vendor-conditioned execution paths
it does not mean the runtime lane is equally strong for every family
it does not mean the strongest analytical sentence automatically becomes a stronger lab consequence sentence

What Review-Grade Bounded Still Means¶

the family is real enough to inspect through benchmark, runtime, grounding, and recommendation surfaces
the family is not weak or fake; it is explicitly narrowed by visible pressure that still survives challenge
the narrower released sentence is part of the product honesty, not a contradiction in the docs

The current lab-consequence evidence is still benchmark-simulated rather than live wet-lab release proof. It earns bounded consequence language because the requested-versus-observed loop is now explicit, not because the repository has already closed real laboratory deployment risk.

Why LFQ Still Narrows At Release Time¶

LFQ now has a full outsider-auditable packet in the narrow sense: public benchmark packages, runtime lane, grounding route, recommendation posture, requested-versus-observed outcome dossier, and assay-worth-it ledger row
the released sentence still remains review-grade bounded because missingness, normalization pressure, and sparse-cohort transfer risk remain visible in its own external review and acceptance surfaces
this is intentional honesty rather than inconsistency: the packet is public, and the narrowing pressure is public too

Why This Matrix Matters More Now¶

the repository can now show several family packets with real outsider-facing review substance, so the release boundary has to be explicit instead of implied
stronger public benchmark and rerun surfaces make overclaiming easier unless one page still names the narrower sentence exactly
consequence pressure now matters enough that a family can look analytically strong and still need a weaker released sentence

Reader Use¶

use this page when one trust page sounds strong and you need the release boundary in one place
use it to distinguish family-packet breadth from released family language
use it before widening workflow language in changelogs, READMEs, or release summaries

Family Transfer Evidence¶

dda
companion package: packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/dda_cross_engine_review_package
generalization report: packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/dda_cross_engine_review_package/cross_package_generalization.json
dia
companion package: packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/dia_matrix_shift_review_package
generalization report: packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/dia_matrix_shift_review_package/cross_package_generalization.json
lfq
companion package: packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/lfq_sparse_contrast_review_package
generalization report: packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/lfq_sparse_contrast_review_package/cross_package_generalization.json
multiplex
companion package: packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/multiplex_channel_stress_review_package
generalization report: packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/multiplex_channel_stress_review_package/cross_package_generalization.json
ptm
companion package: packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/ptm_ambiguity_stress_review_package
generalization report: packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/ptm_ambiguity_stress_review_package/cross_package_generalization.json
targeted
companion package: packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/targeted_carryover_review_package
generalization report: packages/bijux-proteomics-core/benchmark-assets/flagship-public-packages/targeted_carryover_review_package/cross_package_generalization.json

Boundary¶

This matrix does not grant elite language. It only narrows what each workflow family can honestly claim today.