Workflow Literature Audits

bijux-proteomics-knowledge compares curated scientific reading with each shipped benchmark sentence. The audit preserves where the literature:

• still supports the shipped workflow-family sentence
• is aging and needs recheck discipline
• already outruns the benchmark or comparator packet we currently ship
• should slow recommendation posture instead of silently being assumed

Audit Questions

The literature audit stack answers four direct questions independently of whether a Runtime lane executes:

• which citations currently ground this workflow family
• when those citations were last rechecked in the curated registry
• where the literature still outruns the shipped benchmark or comparator proof
• which machine-readable bibliography export an outsider should open first

Public Surfaces

• get_workflow_literature_freshness_audit(...)
• get_workflow_bibliography_export(...)
• get_benchmark_literature_gap_matrix()
• get_comparator_literature_gap_matrix()

These are knowledge surfaces, not runtime or intelligence surfaces. They do not decide execution readiness, recommendation posture, or lab consequence. They keep the scientific reading state inspectable so those downstream packages cannot widen language on stale or selectively remembered grounding.

flowchart LR
    registry["curated citation registry"] --> freshness["freshness audit"]
    registry --> bibliography["bibliography export"]
    bibliography --> benchmark_gap["benchmark gap matrix"]
    bibliography --> comparator_gap["comparator gap matrix"]
    freshness --> grounding["claim-grounding review"]
    benchmark_gap --> grounding
    comparator_gap --> grounding
    grounding --> boundary["bounded downstream input"]
    boundary -. no direct promotion .-> recommendation["recommendation posture"]

Opening Order

Open the stack in this order when you are reviewing one workflow family:

1. open the matching workflow claim-grounding page to see the current repository sentence
2. open the freshness audit to see whether the curated citation set is still recently checked
3. open the machine-readable bibliography export to inspect what specific reading is actually carried
4. open the benchmark and comparator gap matrices to see whether current benchmark pressure still lags the literature

This order matters. A bibliography export without the gap matrices can make the reading look broader than the shipped benchmark packet really is.

What The Freshness Audit Means

Freshness here is the curated repository audit state.

It records:

• the latest recheck date preserved in the citation and literature registries
• whether a stable DOI or URL still exists in that curated audit
• whether the current workflow-family summary is already aging past the shipped literature surface

It does not pretend that runtime code is doing live citation crawling or automated truth discovery from the public web.

Machine-Readable Bibliography Exports

Each workflow family exposes one machine-readable bibliography export:

workflow family	bibliography export
dda	workflow_bibliography:dda
dia	workflow_bibliography:dia
lfq	workflow_bibliography:lfq
multiplex	workflow_bibliography:multiplex
ptm	workflow_bibliography:ptm
targeted	workflow_bibliography:targeted

Each export carries title, DOI or stable URL, publication year, relevance tags, contradiction tags, and freshness state.

This is the durable handoff surface for outside review. The point is that a reader should be able to inspect what the repository is actually citing without rebuilding the reading list from prose footnotes.

What The Gap Matrices Add

The two gap matrices answer different questions:

• get_benchmark_literature_gap_matrix() asks where the shipped benchmark package still under-covers the stronger reading
• get_comparator_literature_gap_matrix() asks where comparator pressure still under-covers the stronger reading

That separation matters because one family can look well cited and still remain weak in benchmark breadth, comparator stress, or both.

Audit Dispositions

Audit result	Required disposition
citations are current and aligned with the benchmark sentence	retain the bounded grounding and its recheck date
literature outruns benchmark or comparator pressure	record the gap and narrow downstream language until the shipped evidence catches up
curated registry is stale	mark review due; do not present freshness as current scientific confirmation
a material source contradicts the shipped sentence	route the contradiction explicitly rather than averaging it into supporting evidence
source text cannot be redistributed	preserve citation identity and licensed metadata without copying prohibited content

Evidence Limits

• that every cited study is equally strong or equally relevant
• that literature support by itself upgrades workflow-family public wording
• that freshness replaces contradiction handling, benchmark pressure, or lab consequence

Best Next Routes

• Open Workflow Claim Grounding when the question is which exact sentence the repository currently thinks is justified.
• Open Workflow Families when the question broadens from one reading surface to the whole family packet.
• Open Workflow Recommendation Challenges when the dispute is no longer about the reading itself but about what the decision layer did with it.

Authority Boundary

The literature audit owns citation freshness, bibliography identity, and the relationship between current reading and evidence gaps. Claim Grounding owns the supported sentence, Core owns benchmark evidence, Runtime owns execution, and Intelligence owns recommendation posture. No literature-audit result can promote those independent authorities by itself.