# SCRIMED Module Registry Status: scrimed-module-registry-active-no-phi Updated: 2026-06-30 Modules: 19 ## Boundary SCRIMED module registry is a no-PHI product architecture and build-control surface. It does not authorize live PHI, autonomous diagnosis, treatment, prescribing, patient outreach, payer submission, EHR writeback, production connector use, clinical validation, certification, reimbursement claims, or customer go-live. ## GO / NO-GO - GO for no-PHI module architecture, synthetic demos, internal planning, benchmark design, diligence packets, and protected-pilot preparation. - NO-GO for live PHI, autonomous clinical action, patient outreach, payer submission, EHR writeback, production connector use, certification claims, clinical validation claims, or customer go-live. ## Modules - SCRIMED ClinicalBench: synthetic-demo-ready; clinical-evaluation; Create SCRIMED's domain-specific clinical benchmark suite for adversarial, missing-data, noisy-note, unit, guideline, citation, bias, and reviewer-gate evaluation. First milestone: Bind ClinicalBench scenarios to each SCRIMED clinical module and require passing synthetic scorecards before buyer-facing expansion. - SCRIMED Scientific Reasoning Engine: architecture-ready; scientific-reasoning; Structure hypothesis, evidence ranking, contradiction detection, guideline comparison, and uncertainty reporting for scientific and clinical research workflows. First milestone: Create a no-PHI research reasoning packet format with claim, evidence, contradiction, confidence, and reviewer fields. - SCRIMED Workflow Planner: protected-pilot-prep; workflow-orchestration; Generate deterministic workflow plans that map task goals to agents, tools, approvals, evidence, rollback, and audit requirements. First milestone: Create a workflow-plan schema with deterministic states, required approvals, evidence bindings, and rollback behavior. - SCRIMED Clinical Judgment Engine: blocked-before-clinical-production; clinical-evaluation; Assist qualified reviewers by structuring evidence, uncertainty, contraindication flags, escalation criteria, and reviewer disposition for clinical judgment support. First milestone: Add a clinical judgment packet schema with explicit not-for-care boundary, evidence sufficiency, and reviewer signoff state. - SCRIMED Evidence Graph: architecture-ready; evidence-intelligence; Represent claims, sources, citations, guidelines, patient-safe synthetic facts, conflicts, provenance, and confidence as a traceable evidence graph. First milestone: Define graph node and edge schemas for claim, source, citation, guideline, conflict, freshness, and review status. - SCRIMED ResearchOps: architecture-ready; scientific-reasoning; Coordinate literature review, evidence ranking, trial matching simulation, contradiction review, and research workflow operations. First milestone: Create a research pipeline state machine from hypothesis to evidence ranking to reviewer packet. - SCRIMED LongTask Runtime: protected-pilot-prep; runtime; Run durable, resumable, review-gated tasks for research, evaluation, workflow planning, ingestion, and audit packet generation. First milestone: Create a durable task contract with checkpoint, pause-for-review, resume, retry, timeout, and dead-letter states. - SCRIMED Clinical Work Graph: architecture-ready; workflow-orchestration; Model clinical and administrative work as nodes, dependencies, evidence, owners, risk labels, approval states, and blocked actions. First milestone: Define a graph schema for work nodes, evidence refs, owner, risk, state, approval, and rollback. - SCRIMED Trust Score: synthetic-demo-ready; trust-safety; Compute module, agent, model, prompt, workflow, and output trust posture from evidence quality, eval performance, reviewer outcomes, drift, incidents, and policy decisions. First milestone: Create a deterministic trust score formula with explainable components and release-gate thresholds. - SCRIMED Continuous Evaluation Platform: protected-pilot-prep; continuous-evaluation; Continuously evaluate modules, agents, prompts, model routes, workflows, RAG, memory, and trust score regressions. First milestone: Add module-version eval mapping and require continuous eval status in release evidence. - SCRIMED Outcome Intelligence: blocked-before-clinical-production; outcome-learning; Analyze approved synthetic, de-identified, or governed outcome signals to understand workflow value, safety, quality, and operational impact. First milestone: Define outcome signal schema with consent/data-use status, de-identification status, evidence link, and reviewer owner. - SCRIMED Knowledge Evolution Engine: architecture-ready; knowledge-memory; Version and evolve approved knowledge, evidence, guidelines, memory, prompts, and module policies through review-gated updates. First milestone: Create knowledge version records with source, owner, policy, eval impact, reviewer, and rollback fields. - SCRIMED Multi-Agent Runtime: protected-pilot-prep; multi-agent-ai; Coordinate planner, specialist, verifier, evidence, QA, safety, cost, and supervisor agents with scoped identities and audit traces. First milestone: Define multi-agent run envelope with planner, specialist, verifier, supervisor, cost, policy, and trace fields. - SCRIMED Multi-Model Router: starter-build-ready; multi-agent-ai; Route tasks across frontier, open, local, edge, and fallback models by cost, risk, latency, quality, privacy, and policy. First milestone: Bind module-level task classes to the existing SCRIMED model router and cost guardrails. - SCRIMED Adaptive Workflow Selector: architecture-ready; workflow-orchestration; Select the safest workflow template based on task type, clinical risk, data completeness, policy, user role, and reviewer capacity. First milestone: Create workflow selection rules with risk gates, missing-data fallbacks, and reviewer queue constraints. - SCRIMED Clinical Memory: blocked-before-clinical-production; knowledge-memory; Store approved, task-scoped clinical context memory with consent, provenance, purpose-of-use, retention, and deletion controls. First milestone: Define a no-PHI memory record with source, purpose, scope, TTL, reviewer, and deletion event fields. - SCRIMED Research Memory: architecture-ready; knowledge-memory; Retain approved research hypotheses, evidence trails, search strategies, trial criteria, contradictions, and reviewer outcomes. First milestone: Create research memory schema with hypothesis, sources, search method, contradictions, reviewer, and update status. - SCRIMED Physician Preference Memory: blocked-before-clinical-production; knowledge-memory; Capture approved clinician workflow preferences for drafting style, routing, documentation support, and non-clinical defaults without overriding clinical judgment. First milestone: Define opt-in preference memory with scope, owner, TTL, version, reset, and audit fields. - SCRIMED Benchmark Studio: synthetic-demo-ready; continuous-evaluation; Give SCRIMED operators a studio for designing, running, comparing, and publishing internal synthetic benchmarks and release evidence. First milestone: Create benchmark definition schema with scenario set, run metadata, score rubric, comparison targets, and release packet export. ## Validation - PASS all-requested-modules-registered: All 19 requested SCRIMED modules must be present in the registry. - PASS human-review-boundary: Every module must preserve human review for protected clinical workflows. - PASS no-direct-llm-database-access: Every module must retain the no direct LLM-to-database invariant. - PASS blocked-production-boundaries: Every module must declare blocked production modes. - PASS evidence-and-audit-required: Every module must require evidence and auditability. ## Next Implementation Step Promote SCRIMED ClinicalBench, Trust Score, Benchmark Studio, and Continuous Evaluation Platform first so every new module has measurable safety and readiness evidence before workflow execution expands.