Underwriting Decision MCP
Demo MCP for insurance underwriting decision support, reading application forms, health questionnaires, and medical exam results to return justified recommendations using deterministic rules.
README
Underwriting Decision MCP
社内デモ用の保険引受判断支援 MCP。申込書・健康告知書・健康診断結果を読み取り、 根拠付きの判断候補を返す。最終的な引受決定は行わず、recommendation は決定論ルール エンジンでのみ決める。
仕様の正本は UNDERWRITING_MCP_IMPLEMENTATION_V2.md。 旧版は UNDERWRITING_MCP_IMPLEMENTATION.md。
Local Demo
make install
make generate-samples
make test
make dev-mcp
APP_MODE=mock ではAWSなしで Case A/B/C fixture を処理できる。実PDFアップロードの
抽出はAWS pipeline用で、mockでは NOT_AVAILABLE_IN_MODE を返す。
V2 Distribution Model
- Cowork on 3P inference: Amazon Bedrock Claude
- Remote MCP: internet-facing HTTPS endpoint, distributed by
managedMcpServers - Plugin: skill-only organization Plugin, no
.mcp.json, no MCP secret - Upload: Plugin script posts PDFs to
/v1/cases/{case_id}/documents/{document_type} - Auth:
AUTH_MODE=jwtfor public MCP; upload tokens are separate short-lived single-use tokens
Other-PC distribution assets live in deploy/cowork. The Plugin package is produced by:
make package-plugin
Data Boundary
This demo is not air-gapped. The intended boundary is that business data is not sent
to Anthropic's inference platform. Cowork runtime download traffic to downloads.claude.ai
is expected. See docs/DATA_BOUNDARY.md.
Main Layout
underwriting_core/: deterministic rules, canonical facts, result assembly, explain, what-ifunderwriting_app/: service layer, job/case models, ports, mock adapters, upload validationunderwriting_mcp/: FastMCP server, JWT auth, Upload APIlambdas/underwriting_workflow/: Textract, Bedrock normalization, final assembly workerplugin/underwriting-review/: skill-only organization Plugindeploy/cowork/: managed configuration templates and installer scriptsinfra/: AWS CDK v2 stackssamples/: fictional demo cases and generated PDFsdocs/: architecture, deployment, data boundary, security, demo notes
Live Smoke
After deployment and token issuance:
export MCP_URL="https://<mcp-host>/mcp"
export MCP_BEARER_TOKEN="<demo-jwt>"
RUN_AWS_LIVE_TESTS=1 make smoke-test
This creates a case, uploads generated PDFs, starts the AWS workflow, polls the result, and checks the expected recommendation.
AWS Cost Modes
The default CDK deployment is a private, production-like demo shape with Interface VPC Endpoints.
For short live demos with only fictional sample forms, use -c deploymentMode=demo-low-cost to
avoid those endpoint-hour charges. In that mode the workflow Lambda runs outside the VPC and the
MCP Fargate task runs in public subnets behind the HTTPS ALB. Tokyo low-cost demos use deterministic
demo extraction because Amazon Textract is not available in ap-northeast-1. See
docs/DEPLOY.md.
判定区分: ELIGIBLE_CANDIDATE / REFER / NOT_ELIGIBLE_CANDIDATE。
すべて架空のデモ用ルールであり、実在保険会社の基準ではない。
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.