Laboratory foundations

Lab owns the boundary where an advisory scientific action becomes an operational assay, incurs material and opportunity cost, and produces an observed consequence. It protects that boundary with readiness, controls, explicit handoff, QC, reconciliation, and refusal.

An assay consequence is therefore more than a proposed experiment. It joins control demands, queue or material burden, operator custody, acceptance rules, and observed outcomes in one reviewable chain. If any link is missing, Lab must retain the plan as advisory, narrow it, or refuse execution.

flowchart LR
    R["advisory recommendation"] --> D["assay design"]
    D --> G["readiness gates"]
    G -->|pass| H["authorized handoff"]
    G -->|fail| F["refuse or revise"]
    H --> E["experiment"]
    E --> O["observation and QC"]
    O --> C["reconciliation"]
    C --> K["new Knowledge evidence"]
    C --> I["new policy feedback"]

Laboratory routes

Question	Guide	Governing record
What does Lab own?	Package overview	design, readiness, handoff, observation, and reconciliation
Is a proposal a lab concern?	Ownership boundary	operational assay consequence rather than upstream policy
Which planning surfaces exist?	Capability map	plans, queues, batches, schedules, handoffs, outcomes
What remains outside?	Scope and non-goals	scientific truth, recommendation ranking, and general runtime execution
What burden does follow-up impose?	Lab consequence	controls, materials, capacity, risk, and opportunity cost
Why did a plan stop?	Workflow refusal handbook	reason code, blocking evidence, and safe next action
How do outcomes change future work?	Outcome learning loops	reconciliation and versioned feedback

The package boundary resolves cases where recommendation policy or generic orchestration is being mislabeled as laboratory ownership.

Readiness protocol

An executable assay requires more than a plausible recommendation:

1. The biological question and measurable endpoint are explicit.
2. The upstream evidence and recommendation records resolve by identity and version.
3. Positive, negative, process, and interpretation controls are sufficient for the stated question.
4. Protocol, materials, samples, instrument requirements, acceptance criteria, and failure handling are complete.
5. Capacity, batching, scheduling, safety, and opportunity-cost constraints are acceptable.
6. Ownership and handoff custody are unambiguous.
7. The outcome can be reconciled back to the requested measurement.

An advisory plan may identify missing items. Only an executable plan has cleared the stronger operational gate.

Consequence, burden, and refusal

The lab consequence route makes downstream cost visible before work is normalized into a queue. Burden includes scarce sample, reagents, instrument time, operator time, controls, competing assays, and the risk of producing an uninterpretable result.

flowchart TD
    A["advisory plan"] --> Q{"question and endpoint clear?"}
    Q -->|no| R["refuse: resolve scientific premise"]
    Q -->|yes| C{"controls and protocol complete?"}
    C -->|no| V["revise design"]
    C -->|yes| B{"burden and capacity acceptable?"}
    B -->|no| N["narrow, queue, or refuse"]
    B -->|yes| E["executable plan"]

Refusal preserves the reason work is unsafe, wasteful, under-specified, or unable to answer the question. It also identifies a valid next action. A lab refusal is not an execution crash and does not invalidate the upstream result; it bounds what may happen next.

This Package Does Not Own

Lab does not establish scientific truth, maintain the evidence graph, rank recommendations, or provide general-purpose execution infrastructure. It owns the stronger operational question: whether a proposed assay can be authorized under its control demands, queue or material burden, safety constraints, and acceptance rules. Knowledge, Intelligence, and Runtime retain their respective authorities throughout that handoff.

Handoff and observation

An authorized handoff binds stable assay, batch, sample, material, protocol, control, and acceptance identities to an operator and schedule. Deviations are recorded rather than absorbed into narrative notes.

Observation records facts: measurements, QC, missingness, deviations, and operational failures. Reconciliation compares those observations with the requested endpoint and classifies the follow-up as supported, weakened, rejected, inconclusive, or requiring another cycle. The outcome learning loop then creates new Knowledge evidence and, when appropriate, a new Intelligence calibration record.

Evolution rules

The domain language stabilizes advisory, executable, readiness, handoff, observation, reconciliation, and consequence. The change principles protect append-only history and authority boundaries when plans or policies evolve. Dependencies and adjacencies, repository fit, and lifecycle overview connect the lab loop to neighboring packages without granting Lab ownership of their scientific, evidential, or execution state.