Dental Clinic Loan Verification MCP Server
Enables automated dental clinic loan verification by combining rule-based ID validation with LLM-powered document analysis and fraud detection. It supports provider-agnostic vision and reasoning tools to assess document consistency, verify credentials like PAN and GST, and generate comprehensive risk narratives.
README
Dental Clinic Loan Verification — FastMCP Server
Provider-Agnostic LLM · Local (Slingshot) → Cloud (Horizon)
Switch LLM Provider — Zero Code Changes
# Use Anthropic Claude (default)
LLM_PROVIDER=anthropic
ANTHROPIC_API_KEY=sk-ant-...
# Use Google Gemini
LLM_PROVIDER=google
GOOGLE_API_KEY=AIza...
| Provider | Default model | Override |
|---|---|---|
| anthropic | claude-sonnet-4-20250514 | LLM_MODEL=claude-opus-4-5 |
| gemini-2.0-flash | LLM_MODEL=gemini-1.5-pro |
Phase 1 — Local Dev with VS Code / Slingshot
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env # add your API key
.vscode/mcp.json is already in the project — VS Code and Slingshot auto-discover it.
Open the folder in VS Code and the tools appear automatically in agent mode.
To switch to Gemini in VS Code, edit .vscode/mcp.json:
"env": { "LLM_PROVIDER": "google", "GOOGLE_API_KEY": "${env:GOOGLE_API_KEY}" }
python demo.py # end-to-end demo
mcp dev server.py # browser tool inspector
Phase 2 — Cloud Deploy on Horizon
# 1. Push to GitHub (.env in .gitignore)
git add server.py llm_tools.py llm_client.py requirements.txt .vscode/mcp.json
git push
# 2. Connect repo at https://horizon.prefect.io
# Horizon auto-detects requirements.txt
# 3. Set secrets in Horizon dashboard:
# LLM_PROVIDER=anthropic
# ANTHROPIC_API_KEY=sk-ant-...
# 4. Get live URL: https://dental-loan-verifier.fastmcp.app/mcp
Switch Slingshot from local → cloud by updating .vscode/mcp.json:
{
"servers": {
"dental-loan-verifier": {
"type": "http",
"url": "https://dental-loan-verifier.fastmcp.app/mcp"
}
}
}
Zero code changes between local and cloud.
File Structure
dental_loan_mcp/
├── server.py ← FastMCP server + rule-based tools
├── llm_tools.py ← LLM sub-agent tools (provider-agnostic)
├── llm_client.py ← Provider switch: Anthropic / Google
├── demo.py ← Local end-to-end demo
├── requirements.txt
├── .env.example ← Copy → .env, add keys
└── .vscode/
└── mcp.json ← Auto-discovered by VS Code + Slingshot
Tool Reference
| Tool | Type | LLM task |
|---|---|---|
doc_validate_document_set |
Rule-based stub | — |
pan_validate |
Rule-based stub → NSDL | — |
fraud_registry_check |
Rule-based stub | — |
aadhaar_verify |
Rule-based stub → UIDAI | — |
dci_credential_verify |
Rule-based stub → DCI | — |
gst_verify |
Rule-based stub → GST | — |
llm_health_check |
LLM | Shows active provider/model |
llm_analyze_document |
LLM Vision | Reads doc image, detects photo gender |
llm_cross_check_documents |
LLM Reasoning | Cross-doc inconsistencies |
llm_assess_name_match |
LLM Language | Indian name judgment |
llm_fraud_reasoning |
LLM Reasoning | Fraud signal clustering |
llm_generate_risk_narrative |
LLM Language | Final report writing |
orchestrate_stage2_verification |
Orchestrator | Full execution plan |
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。