Validation

Validation exercises the configured interfaces and records the outcome. Production readiness depends on validation evidence, not on a self-declared status.

• A connection test proves reachability, authentication and contract version agreement — not that the laboratory system accepts orders.
• Sandbox scenarios prove the full order, specimen, result and exception path against synthetic data.
• Validation evidence is scoped to one environment: a Sandbox run never counts toward Production, and a stale or failed read is treated as not validated.
• Any unresolved requirement remains visible as a production blocker until it is genuinely resolved.
• Native interface validation against a real laboratory system requires that laboratory's approved test endpoint, test patients, signed test plan and clinical sign-off; synthetic tests are never presented as vendor certification.

• Confirm workspace, selected environment, adapter contract/version and responsible laboratory IT owner. Record the current connection or mapping revision without exporting credentials.
• Follow the existing six steps: Laboratory information, Testing capabilities, Fulfillment, Test your setup, Connect your laboratory, Production readiness. Identify the LIS product/version and licensed interface in Step 5; Step 6 remains a separate technical and commercial decision.
• Preflight the approved endpoint/network, catalog orderables, specimen and ordering-party requirements, current immutable profile revision and independent mapping reviewer. Missing tenant scope or installation authority leaves testing blocked; a gateway or Edge probe supplies no approval.
• Run the supported connection check and inspect its actual outcome. Reachability and authentication are prerequisites, not order acceptance.
• In Sandbox, create only synthetic scenarios through the existing workflow. Correlate the order, acknowledgement, specimen/result and exception records in that same environment.
• Resolve missing mappings, unsupported collection/region requirements and exceptions through their canonical workflows, then repeat the affected synthetic scenario. A display label is not evidence of resolution.
• Inspect current evidence after refresh; failed, missing, expired or wrong-path evidence remains incomplete. Step 5 completion and Step 6 readiness are separate conclusions.
• Laboratory connection/configuration changes and deployment validation require authorized Lab Admin or platform-operator actions. Operations contributes IT information, not self-granted Production approval.

• Synthetic Sandbox: the existing fixture journey demonstrates its configured synthetic path. It never certifies a real LIS or counts as Production validation.
• Standalone protocol tests: locally hosted REST/FHIR, HL7/MLLP and SFTP peers check parsing, envelopes and failure handling in source. They do not demonstrate Core installation, hosted dispatch or vendor compatibility.
• Approved real laboratory test: requires the laboratory's explicit endpoint/test identity authorization, signed test plan, exact profile/mapping revision and independent clinical review. No real-system test is established for the candidate vendors.
• Post-live Production diagnostics: fresh, authorized metadata-only reads inspect current credentials/grants, connection and activity state. They are not permission to create a clinical order, inject a result, use a synthetic patient or rerun Sandbox in Production.

• Retain protocol, authorized profile revision, reviewer and proof source alongside safe correlation/idempotency references and the observed test date.
• Reconcile accession/specimen, supported orderables, observation codes/units and partial/final/corrected/amended versions with authorized release status.
• Negative cases include timeouts, lost acknowledgement, wrong environment, revoked credentials, conflicting mappings and duplicate source events. Refusal or reconciliation is not a completed pass.

Keep the safe run/request reference, workspace, environment, selected path, adapter/profile/mapping revision, observed timestamps and order/result correlation together. Compare any expiry supplied by the canonical evidence; do not invent a universal validity duration. Replaced configuration, an expired approval or a failed current read cannot inherit an old pass.

Transport receipt, HL7 commit acknowledgement, application acceptance, accession/specimen acceptance and final or corrected result release are separate milestones. The laboratory reviewer confirms clinical mapping, units, status semantics and release policy; Solia does not infer sign-off from a successful request.

When validation is refused, preserve its safe error and current evidence references, identify the missing owner/approval and escalate through the existing integration workflow. Do not use a generic clinical write or a manual resend to force a pass. Authorized Production, first real LIS certification and second-site zero-source-change reuse remain separate blocked acceptance milestones.

The selected CI verifies 29 Prompt4 candidate wire/draft tests, separate disposable two-buyer 7/7 and integrated 24/24 within 1,632 selected tests. The candidate callback is not one joined canonical transaction. Distinguish pure transformation, loopback, disposable persistence and separately hosted verified acceptance.