Dependencies and adjacencies

Lab sits at the operational end of the evidence chain. Published metadata declares Foundation, Core, and Knowledge as product prerequisites. Its source uses Foundation records throughout, consumes Knowledge evidence for planning and outcome ingestion, and uses Intelligence records in bounded follow-up and review paths. It does not depend on Runtime to define laboratory authority.

flowchart LR
    foundation["Foundation\nidentity and documents"] --> lab["Lab\nreadiness and consequence"]
    core["Core\nscientific request and QC"] --> lab
    knowledge["Knowledge\nevidence and contradictions"] --> lab
    intelligence["Intelligence\nadvisory recommendation"] --> lab
    lab --> observation["observation and consequence"]
    observation --> knowledge
    runtime["Runtime\ngeneral execution"] -. artifact custody only .-> lab

The diagram shows record flow and authority. It is not permission for Lab to rewrite scientific results, evidence bundles, or recommendations as they cross the boundary.

Dependency contract

Dependency	Lab consumes	Ownership that remains upstream
Foundation	assay, batch, program, gate, claim, and document identities; typed outcomes	canonical representation and compatibility
Core	scientific question, measurement requirements, analytical QC, and validation context	calculation, scientific result, and family acceptance
Knowledge	versioned evidence, contradictions, gaps, and observed-evidence ingestion	source custody, relationship, and sufficiency
Intelligence	advisory candidate and follow-up records in bounded review paths	ranking, sensitivity, confidence, and recommendation posture
Pydantic	strict plans, handoffs, observations, and consequence records	readiness policy and human authorization remain explicit Lab contracts

Runtime adjacency

Runtime may execute software that produces an input artifact or transports a lab handoff, but it does not make an assay operationally ready. Lab owns the material, control, protocol, capacity, custody, and answerability checks that separate an advisory plan from an executable handoff.

Runtime fact	Lab decision that still must occur
workflow completed	are the scientific inputs sufficient for assay design?
artifact hash matches	does the handoff contain the required controls, materials, and acceptance criteria?
provider succeeded	is the requested measurement feasible and safe under current capacity?
archive was delivered	did an accountable operator accept custody of exact instructions?

Dependency placement rules

Proposed behavior	Correct owner
proteomics calculation or analytical threshold	Core
source relationship or contradiction resolution	Knowledge
ranking or recommendation policy	Intelligence
generic provider, retry, checkpoint, or run transport	Runtime
assay readiness, scheduling, custody, observation, or requested-versus-observed reconciliation	Lab

Review the edge

Before adding or widening a dependency, verify that:

1. every upstream record retains its owner and immutable identity;
2. advisory and executable plans remain distinct types and states;
3. missing controls, materials, capacity, or authority can produce refusal;
4. observations preserve QC, deviations, failures, and inconclusive outcomes;
5. evidence feedback creates a new Knowledge record rather than rewriting the plan or recommendation.

Continue with lab consequence, outcome learning loops, and dependency governance.