MedRecPro

MedRecPro is a pharmaceutical structured product label (SPL) management platform built with ASP.NET Core. It provides secure access to FDA drug label data through a RESTful API, an AI-powered chat interface, a Model Context Protocol (MCP) server for integration with AI assistants like Claude, and an interactive adverse-event risk dashboard built on a multi-stage table-standardization and risk-statistics pipeline.

Specifications

• HL7 Version: HL7 Dec 2023 https://www.fda.gov/media/84201/download
• Info: https://www.fda.gov/industry/fda-data-standards-advisory-board/structured-product-labeling-resources
• SPL Data Source: https://dailymed.nlm.nih.gov/dailymed/spl-resources-all-drug-labels.cfm
• Orange Book Data Source: https://www.fda.gov/drugs/drug-approvals-and-databases/approved-drug-products-therapeutic-equivalence-evaluations-orange-book

Technology Stack

• Runtime: ASP.NET Core (.NET 8.0 LTS)
• Database: Azure SQL Server (Serverless free tier) with Dapper + Entity Framework Core
• Authentication: Cookie-based auth with Google and Microsoft OAuth providers; JWT bearer tokens for API access; McpBearer JWT scheme for MCP server integration (claims normalized to standard JWT short names: sub, name, email)
• AI Integration: Claude API for natural language query interpretation and synthesis
• MCP Protocol: Model Context Protocol server with OAuth 2.1 (PKCE S256) for Claude.ai connector integration
• Hosting: Azure App Service (Windows, IIS) with Cloudflare CDN/WAF/DNS
• Secrets: Azure Key Vault
• API Documentation: Swagger/OpenAPI
• SPL Rendering: RazorLight templates for SPL XML-to-HTML generation

Solution Architecture

The solution's three web projects are deployed to a single Azure App Service using IIS virtual applications (the console, library, React SPA source, prototypes, and test projects are not deployed as separate apps):

                        Cloudflare (CDN/WAF/DNS)
                                |
                                v
    Azure App Service: "MedRecPro" (Windows, IIS)
   +--------------------------------------------------------+
   |                                                        |
   |  /                site\wwwroot       MedRecProStatic   |
   |  /adverse-events  (MedRecProStatic, React island)      |
   |  /api             site\wwwroot\api   MedRecPro API     |
   |  /mcp             site\wwwroot\mcp   MedRecProMCP      |
   |                                                        |
   +--------------------------------------------------------+
                        |
                        v
               Azure Key Vault
              (medrecprovault)

Virtual Path	Project	Purpose
/	MedRecProStatic	Static site, marketing pages, AI chat UI, OAuth/MCP discovery metadata, adverse-event dashboard host
/adverse-events	MedRecProReact (hosted by MedRecProStatic)	Adverse-event risk dashboard React island
/api	MedRecPro	REST API: SPL parsing, label CRUD, authentication, AI interpret/synthesize, adverse-event dashboard data
/mcp	MedRecProMCP	MCP server: OAuth 2.1 gateway for Claude.ai integration
(CLI)	MedRecProConsole	Standalone bulk import utility (SPL labels, FDA Orange Book, and table standardization)
(library)	MedRecProImportClass	Shared class library: entity models, parsing services, table-standardization pipeline, and EF Core context for SPL and Orange Book import
(SPA source)	MedRecProReact	React + Vite source for the adverse-event dashboard; builds into MedRecProStatic's web root
(prototypes)	MedRecProPrototypes	Standalone HTML/JS prototypes (e.g. the AE dashboard) that seed production UI work
(test)	MedRecProTest	Unit and integration tests

How the Projects Relate

MedRecProStatic is the user-facing front end. Its AI chat interface (/Home/Chat) communicates with the API using a request-interpret-execute-synthesize pattern: user queries are sent to the API's AI endpoints, which use Claude to map natural language to API calls. The static site also serves OAuth/MCP discovery metadata (/.well-known/*) at the domain root on behalf of the MCP server, because the MCP SDK resolves discovery URLs relative to the domain root rather than the /mcp path.

MedRecPro (API) is the core backend. It handles SPL XML parsing and import, label data CRUD, user authentication, AI query interpretation via Claude, database views for navigation, and SPL document rendering via RazorLight templates.

MedRecProMCP is an OAuth 2.1 gateway that exposes MedRecPro API capabilities as MCP tools. When Claude.ai connects, it authenticates users through Google/Microsoft OAuth, resolves upstream identity provider identities to numeric database user IDs (auto-provisioning new users if needed), then forwards authenticated MCP JWTs to the MedRecPro API. It uses JWT tokens, PKCE (S256), Dynamic Client Registration (RFC 7591), and a shared PKSecret for encrypted user ID exchange with the API.

MedRecProReact is the source for the adverse-event risk dashboard — a React + Vite single-page "island". Its Vite build emits a deterministic bundle directly into MedRecProStatic/wwwroot/ae-dashboard, which MedRecProStatic serves at /adverse-events. The dashboard reads only from the API's /api/AdverseEvent surface (AdverseEventController), which in turn queries the materialized AE risk tables produced by the table-standardization pipeline (Stage 5). Current dashboard focuses cover product-level risk, pharmacologic-class SOC correlation, and MedDRA-system-scoped class correlation. See the MedRecProReact README for the dashboard, and the MedRecProImportClass README for the risk-statistics contract.

Repository File Structure

MedRecPro/                          # Root repository
  README.md                         # This file
  .gitignore
  LICENSE.txt

  MedRecPro/                        # ASP.NET Core Web API
    Program.cs                      # App startup, DI, middleware
    MedRecPro.csproj                # .NET 8.0 project file
    appsettings.json                # Base configuration
    appsettings.Development.json    # Local dev overrides
    web.Release.config              # IIS release config
    Controllers/
      ApiControllerBase.cs          # Base controller (route prefix, #if DEBUG directives)
      AuthController.cs             # OAuth login/logout, user info
      UsersController.cs            # User CRUD, activity logs, authentication, MCP user resolution/provisioning
                                      #   [Authorize(Policy = "ApiAccess")] — accepts both cookie auth and McpBearer JWT
                                      #   signup and authenticate endpoints use [AllowAnonymous]
      LabelController.cs            # Label CRUD, views, search, import, AI endpoints
      AiController.cs               # AI interpret/synthesize, conversations, context
      SettingsController.cs         # App info, feature flags, metrics, logs, cache
      AdverseEventController.cs     # Adverse-event dashboard data: products, favorites,
                                      #   triage/forest/quadrant, reverse-lookup, interchange,
                                      #   class SOC correlation, and MedDRA-system-scoped
                                      #   class correlation map/heatmap/cell (gated by AeDashboard flag)
    Service/
      SplImportService.cs           # SPL ZIP file import and parsing orchestration
      SplParsingService.cs          # Core SPL XML parsing
      SplDataService.cs             # Database operations for label data
      SplContextService.cs          # SPL document context management
      SplDocumentRenderingService.cs    # SPL-to-HTML rendering via RazorLight
      SplStructuredBodyRenderingService.cs
      SplSectionRenderingService.cs
      SplIngredientRenderingService.cs
      SplPackageRenderingService.cs
      SplCharacteristicRenderingService.cs
      SplAuthorRenderingService.cs
      SplTextContentRenderingService.cs
      SplRenderingRegistrationService.cs
      TarpitService.cs              # IP tracking, delay calculation, endpoint abuse detection
      ViewRenderService.cs          # Razor view rendering
      ZipImportWorkerService.cs     # Background ZIP import worker
      BackgroudTaskService.cs       # Background task management
      DatabaseKeepAliveService.cs   # Keeps Azure SQL Serverless awake during business hours
      AzureTokenCredentialService.cs
      AzureAppHostTokenCredentialService.cs
      ParsingServices/              # 20+ specialized SPL XML parsers
        SectionParser.cs
        ProductIdentityParser.cs
        PackagingParser.cs
        ... (and more)
      ParsingValidators/            # SPL validation services
    DataAccess/
      RepositoryDataAccess.cs       # Core data access layer
      UserDataAccess.cs             # User-specific queries
      DtoLabelAccess.cs             # Label DTO queries (base)
      DtoLabelAccess-Views.cs       # Database view queries
      DtoLabelAccess-Document.cs    # Document queries
      DtoLabelAccess-Ingredient.cs  # Ingredient queries
      DtoLabelAccess-Organization.cs
      DtoLabelAccess-ProductHierarchy.cs
      DtoLabelAccess-ContentHierarchy.cs
      DtoLabelAccess-BatchLoaders.cs
      DtoLabelAccess-AeDashboard.cs # Adverse-event dashboard queries and derivations
      AeDashboardDerivation.cs      # Pure signal/triage/interchange/correlation math
      AeDashboardFavoriteAccess.cs  # User favorite product persistence
      AeCorrelationPipelineModels.cs # Internal (non-response) correlation pipeline records
      ... (and more)
    Middleware/
      TarpitMiddleware.cs           # Progressive delay for 404 abuse and endpoint rate limiting
    Models/                         # Domain models, DTOs, enums
      Labels.cs                     # Core label entities
      User.cs                       # User model
      Import.cs                     # Import models
      Comparison.cs                 # Label comparison models
      SectionStructure.cs           # Section hierarchy
      DocumentRendering.cs          # Rendering models
      TarpitSettings.cs             # Tarpit configuration (thresholds, delays, monitored endpoints)
      ... (and more)
    Skills/                         # AI skill definitions (markdown prompts for Claude)
      skills.md                     # Master skill index
      selectors.md                  # Query routing rules
      retryPrompt.md                # Retry logic prompt
      labelProductIndication.md     # Indication discovery skill
      equianalgesicConversion.md    # Opioid conversion skill
      product-extraction.md         # Product extraction skill
      pharmacologic-class-matching.md
      interfaces/                   # Modular skill interface definitions
        response-format.md
        synthesis-rules.md
        api/                        # API-specific skill docs
          indication-discovery.md
          label-content.md
          equianalgesic-conversion.md
          pharmacologic-class.md
          product-extraction-api.md
          user-activity.md
          cache-management.md
          session-management.md
          data-rescue.md
          retry-fallback.md
      prompts/                      # AI prompt templates
        product-extraction-prompt.md
        pharmacologic-class-matching-prompt.md
    Views/
      SplTemplates/                 # RazorLight templates for SPL XML rendering
        GenerateSpl.cshtml          # Main SPL generation template
        _Section.cshtml             # Section partial
        _Product.cshtml             # Product partial
        _Ingredient.cshtml          # Ingredient partial
        _Packaging.cshtml           # Packaging partial
        _Author.cshtml              # Author partial
        ... (18 templates total)
      Stylesheets/                  # SPL rendering stylesheets
    Helpers/                        # Utility classes
      ClaimHelper.cs                # Centralized claim extraction (cookie auth + MCP JWT)
      EncryptionHelper.cs           # ID encryption/decryption
      ConnectionStringHelper.cs     # DB connection management
      XmlHelpers.cs                 # XML parsing utilities
      ... (and more)
    Auth/
      BasicAuthenticationHandler.cs # Basic auth handler
    Attributes/                     # Custom validation attributes for SPL fields
    Filters/
      ActivityLogActionFilter.cs    # Request activity logging
      RequireActorAttributeFilter.cs      # Actor-based authorization filter
      RequireUserRoleAttributeFilter.cs   # Role-based authorization filter
    Migrations/                     # EF Core migrations
    Exceptions/
    SQL/                            # Database schema and maintenance scripts
      MedRecPro.sql                 # Full database schema
      MedRecPro_Views.sql           # View definitions
      MedRecPro_Indexes.sql         # Index definitions
      MedRecPro-Deployment.sql      # Deployment scripts
      DbTriggerSetup.sql            # Database triggers
      MedRecPro-Export-Import.ps1   # PowerShell export/import script
      MedRecPro-AzureStatus.sql     # Azure status queries
      MedRecPro-AzureRebuildIndex.sql
      MedRecPro-AzureDisableIndex.sql
      MedRecPro-AzureNuke.sql       # Full database reset (use with caution)
      MedRecPro-AzureOnlineQueryEditorRebuildIndex.sql
      MedRecPro-TableNames.sql
      MedRecPro-TableTruncate.sql
      MedRecPro-TableMissingIndexes.sql
      MedRecPro-TableCreate-OrangeBook.sql  # Orange Book table definitions (7 tables)
      MedRecPro-AzureOrangeBookNuke.sql     # Orange Book targeted truncation
      MedRecPro-Table-tmp_FlattenedAdverseEventCoverageTable.sql  # Stage 5 AE source-row coverage audit
      MedRecPro-Table-tmp_FlattenedAdverseEventTable.sql          # Stage 5 RR-ready AE stats (RR/DNRR/CI + PERSISTED log columns)
      MedRecPro-Table-tmp_FlattenedAdverseEventRiskTable.sql      # Materialized dbo.vw_AeRisk for the dashboard
      MedRecPro-Table-tmp_AeDashboardProductCatalog.sql           # Materialized dbo.vw_AeDashboardProductCatalog (picker)
      MedRecPro-AdverseEvent-Export-Import.ps1  # BCP full-refresh of the AE tables (local -> Azure SQL)

  MedRecProStatic/                  # Static site and AI chat interface
    Program.cs                      # Startup, middleware, OAuth discovery endpoints
    MedRecProStatic.csproj          # .NET 8.0 project file
    web.config                      # IIS config (httpErrors PassThrough, handler isolation)
    appsettings.json
    appsettings.Development.json
    Controllers/
      HomeController.cs             # Index, Terms, Privacy, Chat pages
      AdverseEventDashboardController.cs  # /adverse-events React island host
    Middleware/
      TarpitMiddleware.cs           # Progressive delay for 404 abuse and endpoint rate limiting
    Models/
      PageContent.cs                # Strongly-typed content models
      TarpitSettings.cs             # Tarpit configuration (thresholds, delays, monitored endpoints)
    Services/
      ContentService.cs             # JSON content loader
      TarpitService.cs              # IP tracking, delay calculation, endpoint abuse detection
    Views/
      Home/
        Index.cshtml                # Landing page
        Terms.cshtml                # Terms of Service
        Privacy.cshtml              # Privacy Policy
        Chat.cshtml                 # AI chat interface
      AdverseEventDashboard/
        Index.cshtml                # Layout-free React dashboard host (mounts ae-dashboard bundle)
      Shared/
        _Layout.cshtml              # Master layout
        _Masthead.cshtml            # Shared masthead partial (logo, nav incl. "Insight" -> /adverse-events)
    Content/
      config.json                   # Site config (URLs, branding, version)
      pages.json                    # Page content (home, terms, privacy)
    wwwroot/
      ae-dashboard/                 # Committed Vite build output from MedRecProReact
      css/                          # Stylesheets (incl. masthead.css)
      js/
        site.js                     # Global scripts
        chat/                       # AI chat modules (18 files)
          index.js                  # Main orchestrator
          api-service.js            # API communication
          endpoint-executor.js      # API endpoint execution
          batch-synthesizer.js      # Response synthesis
          checkpoint-manager.js     # State checkpoints
          checkpoint-renderer.js    # Progress UI rendering
          message-renderer.js       # Chat message rendering
          markdown.js               # Markdown-to-HTML
          config.js                 # Chat configuration
          state.js                  # Client state management
          ... (and more)
      lib/                          # Third-party (Bootstrap, jQuery)

  MedRecProMCP/                     # MCP Server (OAuth 2.1 gateway)
    Program.cs                      # Startup, DI, endpoint mappings
    MedRecProMCP.csproj             # .NET 8.0 project file
    server.json                     # MCP registry metadata
    web.config                      # IIS config
    appsettings.json / .Development.json / .Production.json
    Configuration/
      McpServerSettings.cs
      MedRecProApiSettings.cs
      JwtSettings.cs
      OAuthProviderSettings.cs
    Endpoints/
      OAuthEndpoints.cs             # OAuth authorize, token, register, callbacks
      OAuthMetadataEndpoints.cs     # .well-known metadata
    Services/
      McpTokenService.cs            # JWT token generation/validation
      OAuthService.cs               # OAuth flow orchestration
      ClientRegistrationService.cs  # Dynamic Client Registration (RFC 7591)
      PkceService.cs                # PKCE implementation
      FilePersistedCacheService.cs  # File-based persistent cache
      MedRecProApiClient.cs         # HTTP client for API calls
      UserResolutionService.cs      # Resolves upstream IdP email to numeric DB user ID
    Handlers/
      TokenForwardingHandler.cs     # Forwards MCP JWT to API (DelegatingHandler)
    Helpers/
      StringCipher.cs               # AES encryption (copy from API for user ID decryption)
    Tools/
      DrugLabelTools.cs             # MCP tools: drug label search and export
      UserTools.cs                  # MCP tools: user/account operations
    Models/
      AiAgentDtos.cs                # AI integration models
      WorkPlanModels.cs             # Work plan models
    Templates/
      McpDocumentation.html         # Embedded docs page

  MedRecProConsole/                 # Bulk import CLI tool (SPL + Orange Book)
    Program.cs                      # Entry point, interactive menu, CLI argument dispatch
    Services/
      ImportService.cs              # SPL import orchestration
      ImportProgressTracker.cs      # SPL progress tracking
      OrangeBookImportService.cs    # Orange Book import orchestration (ZIP extraction, truncation, progress)
    Models/
      AppSettings.cs
      CommandLineArgs.cs            # CLI args: --orange-book, --nuke, --connection, --auto-quit, --verbose
      ImportParameters.cs
      ImportQueueItem.cs
      ImportResults.cs
      ImportProgressFile.cs
    Helpers/
      ConfigurationHelper.cs
      ConsoleHelper.cs              # Interactive menu (import, orange-book/ob, database/db, help, quit)
      HelpDocumentation.cs

  MedRecProImportClass/             # Shared class library for import operations
    Models/
      OrangeBook.cs                 # Orange Book entity classes (Applicant, Product, Patent, Exclusivity, junctions)
      ... (SPL models)
    Service/
      ParsingServices/
        OrangeBookProductParsingService.cs  # Orange Book products.txt parsing, batch upserts, entity matching
        ... (20+ SPL parsers)
      TransformationServices/             # SPL table-standardization pipeline (Stage 0 -> Stage 5)
        TableStandardizationServiceCollectionExtensions.cs  # AddTableStandardization(...) DI graph
        ColumnStandardizationService.cs   # Stage 3.25 column standardization (SOC normalization, Phase 2 content)
        AeParameterCategoryDictionaryService.cs  # Scoped service (1,189 entries) resolving NULL ParameterCategory → canonical SOC
        IAeParameterCategoryDictionaryService.cs # Interface for AE ParameterCategory dictionary lookup
        AdverseEventTableFlattening/      # Stage 5 AE denormalization (RR/DNRR/CI) + RelativeRiskCalculator
        ... (parsers, validators, dictionaries — see MedRecProImportClass/README.md)
    TableStandards/                 # Normalization rules, column contracts, and table-type definitions
      normalization-rules.md        # Deterministic Tier 1 rules + ML.NET Tier 2 guidance
      column-contracts.md           # Per-TableCategory column contracts
      table-types.md                # TableCategory classification decision tree
    Context/
      ApplicationDbContext.cs       # EF Core context (auto-registers OrangeBook entities via reflection)

  MedRecProReact/                   # React + Vite source for the adverse-event dashboard
    index.html                      # Vite entry (mounts #aeDashboardApp / #root)
    vite.config.js                  # base /ae-dashboard/, builds into MedRecProStatic/wwwroot/ae-dashboard
    package.json                    # React 19, Vite 8, Vitest 4, ESLint 10
    src/
      App.jsx                       # Dashboard shell (product + class + system focus, all panels)
      api/                          # /api/AdverseEvent client + dev/prod base resolution
      lib/ hooks/ components/       # Normalizers, scales, hooks, charts (forest/quadrant/correlation/system)
      test/                         # Vitest specs (see MedRecProReact/README.md)

  MedRecProTest/                    # Unit and integration tests
    SplImportServiceTests.cs
    ProductRenderingServiceTests.cs
    ComparisonServiceTests.cs
    ColumnStandardizationServiceTests.cs  # Column standardization + AE dictionary integration tests
    AeParameterCategoryDictionaryServiceTests.cs  # AE dictionary service unit tests (17 tests)
    UserDataAccessTests.cs
    LogActivityAsyncTests.cs
    StringCipherTests.cs
    ResolveMcpUserTests.cs          # MCP user resolution and auto-provisioning tests
    TarpitServiceTests.cs           # Tarpit service unit tests (404 tracking + endpoint abuse)
    TarpitMiddlewareTests.cs        # Tarpit middleware integration tests
    AdverseEventControllerTests.cs  # AE dashboard controller routing, auth, validation, feature gating
    AeDashboardDataAccessTests.cs   # AE dashboard query + correlation data-access tests
    AeDashboardDerivationTests.cs   # AE signal/triage/interchange/correlation derivation math tests

API Endpoints Summary

All API endpoints are accessed under /api in production (IIS virtual application). Controllers use #if DEBUG directives to handle the path prefix difference between local development (/api/[controller]) and production where IIS strips the /api prefix.

Authentication (/api/Auth)

Method	Route	Description
GET	login/{provider}	Start OAuth flow (Google or Microsoft)
GET	external-logincallback	OAuth callback handler
GET	user	Get current authenticated user info
POST	logout	Log out current user
POST	token-placeholder	Token exchange
GET	login	Login page
GET	loginfailure	Login failure handler
GET	lockout	Account lockout handler
GET	accessdenied	Access denied handler

Users (/api/Users)

Method	Route	Description
GET	me	Get current user profile
GET	{encryptedUserId}	Get user by encrypted ID
GET	byemail	Get user by email
POST	signup	Create new user account
POST	authenticate	Authenticate user
PUT	{encryptedUserId}/profile	Update user profile
DELETE	{encryptedUserId}	Delete user account
PUT	admin-update	Administrative user update
POST	rotate-password	Rotate user password
GET	user/{encryptedUserId}/activity	Get user activity log
GET	user/{encryptedUserId}/activity/daterange	Get activity within date range
GET	endpoint-stats	Get endpoint performance statistics
POST	resolve-mcp	Resolve email to encrypted user ID (McpBearer auth; auto-provisions new users)

Labels (/api/Label)

The main data controller with 40+ endpoints covering navigation views, search, CRUD, import, rendering, and AI features.

Navigation & Search Views:

Method	Route	Description
GET	product/search	Search products
GET	product/related	Related products
GET	product/latest	Latest product labels
GET	product/latest/details	Latest product label details
GET	product/indications	Product indications search
GET	ingredient/search	Search by ingredient (active/inactive)
GET	ingredient/summaries	Ingredient summary list
GET	ingredient/active/summaries	Active ingredients only
GET	ingredient/inactive/summaries	Inactive ingredients only
GET	ingredient/advanced	Advanced ingredient search
GET	ingredient/by-application	Ingredients by application number
GET	ingredient/related	Related ingredients
GET	labeler/search	Search by manufacturer/labeler
GET	labeler/summaries	Labeler summary list
GET	ndc/search	Search by NDC code
GET	ndc/package/search	Search by NDC package code
GET	application-number/search	Search by application number (NDA/ANDA)
GET	application-number/summaries	Application number summaries
GET	pharmacologic-class/search	Search by pharmacologic class
GET	pharmacologic-class/hierarchy	Pharmacologic class hierarchy
GET	pharmacologic-class/summaries	Pharmacologic class summaries
GET	section/search	Search by LOINC section code
GET	section/summaries	Section summaries
GET	document/navigation	Document navigation tree
GET	document/version-history/{setGuidOrDocumentGuid}	Document version history

Label Content & Rendering:

Method	Route	Description
GET	section/content/{documentGuid}	Get section content for a document
GET	markdown/sections/{documentGuid}	Get label sections as markdown
GET	markdown/export/{documentGuid}	Export full label as markdown
GET	markdown/download/{documentGuid}	Download label markdown file
GET	markdown/display/{documentGuid}	Render label as HTML from markdown
GET	generate/{documentGuid}/{minify}	Generate updated SPL XML
GET	original/{documentGuid}/{minify}	Get original SPL XML
GET	single/{documentGuid}	Get single label details
GET	complete/{pageNumber?}/{pageSize?}	Paginated complete label list

Drug Safety:

Method	Route	Description
GET	drug-safety/dea-schedule	DEA schedule classification

AI-Powered Endpoints:

Method	Route	Description
GET	extract-product	AI-powered product extraction from text
GET	comparison/analysis/{documentGuid}	Get comparison analysis
POST	comparison/analysis/{documentGuid}	Start AI comparison analysis
GET	comparison/progress/{operationId}	Check comparison progress

CRUD & Import:

Method	Route	Description
GET	{menuSelection}/{encryptedId}	Get single entity by type
POST	{menuSelection}	Create entity by type
PUT	{menuSelection}/{encryptedId}	Update entity by type
DELETE	{menuSelection}/{encryptedId}	Delete entity by type
POST	import	Bulk SPL ZIP import
GET	import/progress/{operationId}	Check import progress

Reference:

Method	Route	Description
GET	guide	API usage guide
GET	inventory/summary	Database inventory overview
GET	sectionMenu	Available section menu items
GET	{menuSelection}/documentation	Documentation for a data type

AI (/api/Ai)

Method	Route	Description
GET	context	Get AI context (auth status, demo mode, data counts)
POST	interpret	Interpret natural language query into API endpoint specs
POST	synthesize	Synthesize API results into human-readable response
GET	chat	Convenience endpoint for simple queries
POST	conversations	Create new conversation
GET	conversations/{conversationId}	Get conversation
GET	conversations/{conversationId}/history	Get conversation history
DELETE	conversations/{conversationId}	Delete conversation
GET	conversations/stats	Get conversation statistics
POST	retry	Retry last AI operation

Settings (/api/Settings)

Method	Route	Description
GET	demomode	Check demo mode status
GET	info	Application info
GET	features	Feature flags
GET	database-limits	Database limits
GET	metrics/database-cost	Azure SQL free tier usage and cost projections
POST	clearmanagedcache	Clear managed cache
GET	logs	Activity logs
GET	logs/statistics	Log statistics
GET	logs/categories	Log categories
GET	logs/by-date	Logs filtered by date
GET	logs/by-category	Logs filtered by category
GET	logs/by-user	Logs filtered by user
GET	logs/users	Users with log entries
GET	test/app-credential	Test Azure credentials
GET	test/app-metrics-pipeline	Test metrics pipeline

Adverse Event Dashboard (/api/AdverseEvent)

Backs the React adverse-event risk dashboard. The whole controller is gated by the FeatureFlags:AeDashboard:Enabled flag (returns 503 when disabled). Reads are anonymous (favorite state is enriched for authenticated users); favorite writes require ApiAccess. Class-picker and MedDRA system-picker responses expose pagination/aggregate totals via the X-Page-Number, X-Page-Size, X-Total-Count, and X-Chartable-Count headers.

Method	Route	Description
GET	products	Dashboard product list with KPI/coverage data (paged)
GET	products/catalog	Slim cached product catalog for the picker
GET	products/count	Distinct product inventory count
GET	products/favorites	Authenticated user's favorite products
PUT	products/{documentGuid}/favorite	Add a favorite (idempotent, 204)
DELETE	products/{documentGuid}/favorite	Remove a favorite (idempotent, 204)
GET	products/{documentGuid}/triage	Tiered triage signals for one product
GET	products/{documentGuid}/forest	Forest-plot payload for one product
GET	products/{documentGuid}/quadrant	Risk-vs-precision quadrant payload
GET	reverse-lookup	Products reporting one or more exact AE terms
GET	interchange	Comparator-aware two-product therapeutic interchange comparison
GET	correlation/classes	Pharmacologic classes with AE data (class picker)
GET	correlation	SOC × SOC correlation map for one class
GET	correlation/heatmap	Sparse SOC × drug RR heatmap for one class
GET	correlation/cell	Per-drug drill-down for one correlation cell
GET	correlation/systems	MedDRA System Organ Classes with AE rows (system picker)
GET	correlation/systems/map	Selected-system pharmacologic-class × pharmacologic-class correlation map
GET	correlation/systems/heatmap	Selected-system pharmacologic-class × drug RR heatmap
GET	correlation/systems/cell	Per-term drill-down for one selected-system class-pair cell

MCP Server (/mcp)

The MCP server exposes its own endpoints. See the MedRecProMCP README for full details.

Method	Route	Description
POST	/mcp	MCP Streamable HTTP transport (JSON-RPC)
GET	/mcp/health	Health check
GET	/mcp/docs	HTML documentation page
GET	/mcp/.well-known/oauth-protected-resource	Protected Resource Metadata (RFC 9728)
GET	/mcp/.well-known/oauth-authorization-server	Authorization Server Metadata (RFC 8414)
GET	/mcp/oauth/authorize	OAuth authorization endpoint
POST	/mcp/oauth/token	Token exchange endpoint
POST	/mcp/oauth/register	Dynamic Client Registration (RFC 7591)
GET	/mcp/oauth/callback/google	Google OAuth callback
GET	/mcp/oauth/callback/microsoft	Microsoft OAuth callback

MedRecProStatic and MCP Relationship

MedRecProStatic serves the OAuth/MCP discovery metadata at the domain root because the MCP SDK resolves /.well-known/* relative to the domain, not the MCP endpoint path. When Claude connects to https://www.medrecpro.com/mcp, the SDK looks for discovery at https://www.medrecpro.com/.well-known/oauth-protected-resource and /.well-known/oauth-authorization-server.

These endpoints are registered directly in MedRecProStatic's Program.cs as static JSON responses. Attempts to redirect from the root site to /mcp/.well-known/* failed because 302 redirects cause the MCP SDK to derive the wrong resource URI, and reverse proxying through Cloudflare triggers Bot Fight Mode (403 errors).

MedRecProStatic also has a critical web.config setting (httpErrors existingResponse="PassThrough") placed outside the <location> element so it is inherited by the MCP and API virtual applications. Without this, IIS replaces 401 responses with HTML error pages, breaking the MCP OAuth challenge flow.

Database Schema and SQL Scripts

Database schema definitions and maintenance scripts are maintained in MedRecPro/SQL/. These are the authoritative source for schema updates, view definitions, and index management.

Script	Purpose
MedRecPro.sql	Full database schema (tables, constraints, relationships)
MedRecPro_Views.sql	View definitions used by navigation and search endpoints
MedRecPro_Indexes.sql	Index definitions for query performance
MedRecPro-Deployment.sql	Deployment-time schema updates
DbTriggerSetup.sql	Database trigger configuration
MedRecPro-Export-Import.ps1	PowerShell script for database export/import
MedRecPro-AzureStatus.sql	Azure SQL status and diagnostics queries
MedRecPro-AzureRebuildIndex.sql	Index rebuild for Azure SQL
MedRecPro-AzureDisableIndex.sql	Disable indexes during bulk operations
MedRecPro-AzureOnlineQueryEditorRebuildIndex.sql	Index rebuild via Azure Query Editor
MedRecPro-AzureNuke.sql	Full database reset (destructive)
MedRecPro-TableNames.sql	List all table names
MedRecPro-TableTruncate.sql	Truncate tables for reimport
MedRecPro-TableMissingIndexes.sql	Identify missing indexes
MedRecPro-TableCreate-OrangeBook.sql	Orange Book table definitions (7 tables, indexes, extended properties)
MedRecPro-AzureOrangeBookNuke.sql	Targeted Orange Book truncation with safety preview mode
MedRecPro-Table-tmp_FlattenedAdverseEventCoverageTable.sql	Stage 5 AE source-row coverage / non-RR audit table
MedRecPro-Table-tmp_FlattenedAdverseEventTable.sql	Stage 5 RR-ready AE statistics (RR/DNRR/CI + PERSISTED log columns)
MedRecPro-Table-tmp_FlattenedAdverseEventRiskTable.sql	Materialization of dbo.vw_AeRisk for the dashboard
MedRecPro-Table-tmp_AeDashboardProductCatalog.sql	Materialization of dbo.vw_AeDashboardProductCatalog (picker)
MedRecPro-AdverseEvent-Export-Import.ps1	BCP full-refresh of the AE tables (local SQL Server → Azure SQL, truncate-then-import)

When updating database schemas or views, modify the scripts in MedRecPro/SQL/ and run them against the target database. The MedRecPro_Views.sql file is particularly important as the navigation view queries (ingredient search, labeler search, pharmacologic class hierarchy, etc.) are defined there and power many of the API search endpoints. The adverse-event dashboard is backed by the dbo.vw_AeRisk, dbo.vw_AeDrugSummary, and dbo.vw_AeDashboardProductCatalog views (also in MedRecPro_Views.sql), the last two materialized into tmp_ tables by the Stage 5 pipeline so the dashboard reads without runtime statistics.

AI Skills System

The API includes an agentic AI layer that enables natural language interaction with pharmaceutical data. The system follows a request-interpret-execute-synthesize pattern:

1. User submits a natural language query to POST /api/Ai/interpret
2. Claude interprets the query and returns API endpoint specifications
3. The client executes the specified API endpoints
4. Results are sent to POST /api/Ai/synthesize
5. Claude produces a human-readable response with suggested follow-ups

AI skills are defined as markdown prompt files in MedRecPro/Skills/. Key skills include:

• Indication Discovery - Find drugs by indication/use case
• Equianalgesic Conversion - Opioid dose conversion calculations
• Product Extraction - AI-powered extraction of product details from text
• Pharmacologic Class Matching - Map drugs to pharmacologic classifications
• Label Content - Retrieve and synthesize label sections
• Data Rescue - Fallback strategies for missing or incomplete data

Adverse Event Risk Dashboard

The platform turns the free-text adverse-event tables buried in SPL labels into comparable risk statistics and exposes them as an interactive dashboard at /adverse-events.

The data flow spans three projects:

1. MedRecProImportClass / MedRecProConsole — the SPL table-standardization pipeline parses heterogeneous label tables into a uniform analytical schema and, in Stage 5, pre-computes Relative Risk (RR), Dose-Normalized RR (DNRR), and 95% confidence intervals per adverse-event row. Results are materialized into the tmp_FlattenedAdverseEvent* tables. See the MedRecProImportClass README for the full pipeline and the statistical contract.
2. MedRecPro (API) — AdverseEventController (/api/AdverseEvent) serves dashboard-ready, encrypted-ID payloads from the materialized risk views: per-product triage/forest/quadrant, symptom reverse lookup, comparator-aware two-product therapeutic interchange, pharmacologic-class SOC × SOC correlation maps/heatmaps/cell drill-downs, and MedDRA-system-scoped class × class / class × drug correlation views. The feature is gated by FeatureFlags:AeDashboard:Enabled.
3. MedRecProReact — a React + Vite single-page island that renders the product, class, and By System dashboard focuses. Its build output is committed into MedRecProStatic/wwwroot/ae-dashboard and served by MedRecProStatic. See the MedRecProReact README.

Because the observation units for correlation views are intentionally narrow — a single drug within a class for SOC × SOC maps, or shared selected-system terms for By System class-pair maps — sample sizes can be small. The dashboard is deliberately honesty-first: below-floor cells are suppressed rather than fabricated, non-renderable system matrices return warnings, the By System focus is single-system by design, and comparator-mixed payloads are explicitly flagged. The displayed figures are bounded by what each label discloses and what the parser can extract — absence of a signal is not evidence of its absence in practice.

FDA Orange Book Integration

The platform imports and cross-references data from the FDA's Approved Drug Products with Therapeutic Equivalence Evaluations (Orange Book), linking FDA approval records to existing SPL label data.

Orange Book Database Schema

Seven normalized tables store Orange Book data, with three junction tables linking to existing SPL entities:

Table	Purpose
OrangeBookApplicant	Pharmaceutical companies holding FDA approvals
OrangeBookProduct	Drug products (natural key: ApplType + ApplNo + ProductNo)
OrangeBookPatent	Patent records per product with expiration dates
OrangeBookExclusivity	Marketing exclusivity periods (NCE, ODE, RTO, etc.)
OrangeBookProductMarketingCategory	Junction: OB Product → SPL MarketingCategory (by application number)
OrangeBookProductIngredientSubstance	Junction: OB Product → SPL IngredientSubstance
OrangeBookApplicantOrganization	Junction: OB Applicant → SPL Organization

No foreign key constraints are enforced; relationships are managed by the import module. The main nuke script (MedRecPro-AzureNuke.sql) excludes Orange Book tables, which have their own dedicated truncation script.

Orange Book Import Process

The console application (MedRecProConsole) imports Orange Book data from FDA-published ZIP archives containing tilde-delimited text files. The import pipeline:

1. Extract products.txt from the ZIP archive
2. Parse tilde-delimited rows (14 columns) into normalized entities
3. Upsert applicants and products in batches (5,000 rows per batch)
4. Match applicants to SPL organizations using a two-tier strategy:
   • Tier 1 (Normalized Exact): Strips corporate suffixes (INC, LLC, CORP, LTD, GMBH, etc.), punctuation, and whitespace; case-insensitive exact match
   • Tier 2 (Token Similarity): Two-pass fuzzy matching using Jaccard/containment scoring with a 0.67 threshold; strips pharma noise words on the second pass only when sufficient tokens remain
5. Match ingredients to SPL IngredientSubstance records (semicolon-delimited ingredient lists parsed and matched individually)
6. Match products to SPL MarketingCategory records via application number

The import is idempotent (upsert-based), so re-running is safe without crash recovery queues. Real-time multi-phase progress is displayed via Spectre.Console.

Orange Book CLI Usage

Interactive mode:

cd MedRecProConsole
dotnet run
# Select "orange-book" or "ob" from the menu

Unattended mode:

MedRecProConsole.exe --orange-book "path/to/EOBZIP_2026_01.zip" --nuke --auto-quit

Argument	Description
--orange-book <path>	Path to Orange Book ZIP file
--nuke	Truncate all Orange Book tables before import
--connection <name>	Database connection name from appsettings.json
--auto-quit	Exit after completion
--verbose	Enable debug logging

Getting Started

Prerequisites

• .NET 8.0 SDK (LTS)
• SQL Server (local) or Azure SQL Database
• Visual Studio 2022 or VS Code

1. Clone the repository

git clone <repo-url>
cd MedRecPro

2. Configure the API

Create user secrets for the MedRecPro API project:

cd MedRecPro
dotnet user-secrets init
dotnet user-secrets set "Dev:DB:Connection" "Server=localhost;Database=MedRecProDB;User Id=sa;Password=your-password;"
dotnet user-secrets set "Security:DB:PKSecret" "your-encryption-key"
dotnet user-secrets set "Jwt:Key" "your-jwt-signing-key-min-32-chars"
dotnet user-secrets set "Jwt:Issuer" "MedRecPro"
dotnet user-secrets set "Jwt:Audience" "MedRecUsers"
dotnet user-secrets set "Authentication:Google:ClientId" "your-google-client-id"
dotnet user-secrets set "Authentication:Google:ClientSecret" "your-google-client-secret"
dotnet user-secrets set "Authentication:Microsoft:ClientId" "your-microsoft-app-id"
dotnet user-secrets set "Authentication:Microsoft:ClientSecret:Dev" "your-microsoft-secret"
dotnet user-secrets set "Authentication:Microsoft:TenantId" "your-tenant-id"
dotnet user-secrets set "ClaudeApiSettings:ApiKey" "your-claude-api-key"

3. Set up the database

Run the schema script from MedRecPro/SQL/MedRecPro.sql against your SQL Server instance, then apply views and indexes:

# Apply schema, views, and indexes in order
sqlcmd -S localhost -d MedRecProDB -i MedRecPro/SQL/MedRecPro.sql
sqlcmd -S localhost -d MedRecProDB -i MedRecPro/SQL/MedRecPro_Views.sql
sqlcmd -S localhost -d MedRecProDB -i MedRecPro/SQL/MedRecPro_Indexes.sql

Or run EF Core migrations:

cd MedRecPro
dotnet ef database update

4. Run the projects

# Terminal 1: API (port 5093)
cd MedRecPro
dotnet run

# Terminal 2: Static site (port 5001)
cd MedRecProStatic
dotnet run

# Terminal 3: MCP server (port 5233, optional)
cd MedRecProMCP
dotnet run

# Terminal 4: Adverse-event dashboard (Vite dev server, port 50346, optional)
cd MedRecProReact
npm install
npm run dev

The MVC-hosted dashboard at http://localhost:5001/adverse-events is served from the committed Vite bundle; the Vite dev server above is only for iterating on the React source. After changing React source, run npm run build and commit the regenerated MedRecProStatic/wwwroot/ae-dashboard assets (the .NET build does not run Vite).

5. Import data

Upload SPL ZIP files through the API import endpoint or use the console importer:

cd MedRecProConsole
dotnet run -- --help

SPL ZIP files can be downloaded from the DailyMed SPL Resources page.

Orange Book ZIP files can be downloaded from the FDA Orange Book Data Files page. See the FDA Orange Book Integration section for import details.

Setup Pitfalls and Fixes

IIS Virtual Application Path Stripping

IIS strips the virtual application prefix from requests before forwarding to ASP.NET Core. A request to /api/Label/search arrives at Kestrel as /Label/search. All controllers use #if DEBUG compiler directives to handle this:

#if DEBUG
[Route("api/[controller]")]   // Local: full path
#else
[Route("[controller]")]        // Azure: IIS strips /api prefix
#endif

The same pattern applies to MCP routes and Swagger paths.

Cloudflare + Azure App Service Managed Certificates

Azure App Service Managed Certificates do not work with Cloudflare proxy enabled. Use Cloudflare Origin Certificates instead:

1. Create an Origin Certificate in Cloudflare (SSL/TLS > Origin Server)
2. Convert to PFX: openssl pkcs12 -export -out origin.pfx -inkey key.pem -in cert.pem
3. Upload to Azure App Service > Certificates
4. Bind to custom domains with SNI SSL

OAuth Redirect URI Prefix

Production redirect URIs must include the /api/ prefix:

• Correct: https://your-domain.com/api/signin-google
• Wrong: https://your-domain.com/signin-google

Cloudflare Bot Blocking

Cloudflare has multiple independent bot-blocking systems that can interfere with MCP and server-to-server calls:

• "Block AI Bots" (WAF managed rule) - Must allow Claude-User via AI Crawl Control settings
• "Bot Fight Mode" - Blocks requests from hosting provider IPs. Cannot be bypassed with WAF rules. Whitelist Azure App Service outbound IPs via IP Access Rules (Security > WAF > Tools)
• Always set a User-Agent header on outbound HttpClients to avoid bot detection

Azure SQL Serverless Cold Starts

Azure SQL Serverless auto-pauses after inactivity. Resuming takes 30-60 seconds, which can exceed default timeouts. Mitigations:

• DatabaseKeepAliveService pings the database with SELECT 1 every 14 minutes during business hours (Mon-Fri, 8 AM - 8 PM Eastern) to prevent auto-pause. Each ping cycle includes 3 retry attempts with escalating delays (10s, 30s, 60s) and a 90-second connect timeout to accommodate cold resume. Configured via the DatabaseKeepAlive section in appsettings.json
• EnableRetryOnFailure() is configured on the EF Core DbContext with 3 retries and a 60-second command timeout to handle transient failures across all application database operations
• Increase the auto-pause delay during active development

Key Vault Secret Naming

Key Vault uses -- (double dash) as separator; ASP.NET Core configuration uses : (colon). The framework maps between them automatically:

• Key Vault: Authentication--Google--ClientId
• Config: Authentication:Google:ClientId

Ensure no extra characters (trailing commas, quotes) in Key Vault secret values.

IIS httpErrors PassThrough

Without <httpErrors existingResponse="PassThrough" /> in the root site's web.config, IIS replaces HTTP 401 responses with HTML error pages. This breaks OAuth challenge flows where the WWW-Authenticate header must reach the client. This setting must be in the root site (MedRecProStatic) because child virtual applications inherit it.

JWT Multi-Issuer/Audience

When both the API and MCP server issue JWT tokens with different issuer/audience values, the JWT Bearer handler must accept both using ValidIssuers and ValidAudiences arrays instead of the singular properties.

Azure SQL Free Tier Monitoring

The API includes built-in monitoring for Azure SQL Database's serverless free tier (100,000 vCore seconds/month). The AzureSqlMetricsService queries Azure Monitor Metrics to track consumption, project monthly costs, and recommend throttling levels. See the GET /api/Settings/metrics/database-cost endpoint.

Security Configuration

Security settings should be stored in user secrets (development) or Azure Key Vault (production):

{
  "Authentication:Google:ClientId": "your-google-client-id.apps.googleusercontent.com",
  "Authentication:Google:ClientSecret": "your-google-client-secret",
  "Authentication:Microsoft:ClientId": "your-microsoft-app-id",
  "Authentication:Microsoft:ClientSecret:Dev": "your-microsoft-secret",
  "Authentication:Microsoft:ClientSecret:Prod": "your-microsoft-secret",
  "Authentication:Microsoft:TenantId": "your-tenant-id",
  "Security:DB:PKSecret": "your-encryption-key",
  "Jwt:Key": "your-jwt-signing-key",
  "Jwt:Issuer": "MedRecPro",
  "Jwt:Audience": "MedRecUsers",
  "Jwt:ExpirationMinutes": 60,
  "ClaudeApiSettings:ApiKey": "your-claude-api-key",
  "Dev:DB:Connection": "your-dev-connection-string",
  "Prod:DB:Connection": "your-prod-connection-string"
}

Changing Security:DB:PKSecret will break all existing encrypted URLs, favorites, and bookmarks.

Production Deployment

See the detailed deployment guides in each project's README:

• MedRecProMCP README - MCP server setup, OAuth provider config, Cloudflare rules, Claude.ai connector integration, and troubleshooting
• MedRecProStatic README - Static site deployment, content management, and IIS configuration

Deployment Checklist

1. Publish each project to its virtual application path on Azure App Service
2. Verify Azure Key Vault secrets are configured
3. Purge Cloudflare cache after deployment
4. Test authentication flows (Google and Microsoft OAuth)
5. Verify Swagger UI loads at /api/swagger/index.html
6. Verify MCP health check at /mcp/health
7. Test AI chat at the static site

License

See the LICENSE.txt file for details.