# SCRIMED Work & Intelligence Platform Brief

Status: scrimed-work-intelligence-platform-active-synthetic-no-phi
Updated: 2026-07-13
Sessions: 2
Agents: 11
Tools: 7
Providers: 4

## Boundary
SCRIMED Work & Intelligence Platform is a synthetic/no-PHI, verification-first control plane for workspace sessions, model/tool routing, agent orchestration, context retrieval, artifacts, schedules, voice simulation, learning loops, governance, auditability, rollback, and value telemetry. It does not authorize live PHI, autonomous clinical care, diagnosis, treatment, prescribing, patient outreach, payer submission, EHR writeback, final imaging interpretation, production connector approval, certification claims, customer go-live, or external model calls.

## Control Plane
- Workspace sessions require a Definition-of-Done contract before planning.
- Autonomy follows verification strength, reversibility, evidence quality, privacy sensitivity, and clinical/financial consequence.
- Consequential tools remain disabled and approval-gated by default.
- Protected write routes require AAL2 governance identity, tenant role authorization, server runtime token, idempotency, and the durable-store feature flag.
- Retrieved context is treated as untrusted data with citations, tenant metadata, and confidence scores.
- Verification blocks completion when citations, scope, policy, PHI checks, rollback, budget, loop, or approval criteria fail.
- Lifecycle transitions use authoritative durable state, append-only history, independent reviewer checks, and tenant-scoped idempotency.

## Feature Flags
- workEnabled: true
- multiAgentEnabled: true
- modelRouterEnabled: true
- contextEngineEnabled: true
- artifactEngineEnabled: true
- schedulesEnabled: false
- voiceSimulationEnabled: true
- learningLoopEnabled: true
- consequentialActionsEnabled: false

## Schedules
- Morning oncology review preparation: disabled=true, simulationOnly=true
- Hourly critical-lab monitoring simulation: disabled=true, simulationOnly=true
- Nightly risk-score refresh preview: disabled=true, simulationOnly=true
- Weekly payer report preparation: disabled=true, simulationOnly=true
- Monthly executive dashboard generation: disabled=true, simulationOnly=true

## Production Hardening Gate
- Status: operator_action_required
- Strict non-production smoke eligible: false
- Canary eligible: false
- Evidence-ready gates: 11/14
- Operator-required gates: 3

## Next Production-Hardening Step
Apply and verify the SCRIMED Work completion-queue migration in the approved no-PHI target, then use the AAL2 tenant-admin control to retain verified internal-only completion evidence without granting external authority.