Carrier Accounting MCP
Automates ingestion, normalization, validation, review, and posting of insurance carrier statements to Applied Epic, with safety features like trial mode and human review queues.
README
Carrier Accounting MCP
Insurance Carrier Accounting Automation for Snellings Walters Insurance
Automates the full carrier statement lifecycle: ingest → normalize → validate → review → post to Applied Epic.
Quick Start
# 1. Clone and setup
git clone <repo>
cd carrier-accounting-mcp
pip install -r requirements.txt
playwright install chromium
# 2. Configure
cp .env.example .env
# Edit .env with your BigQuery and Epic credentials
# 3. Create BigQuery tables
bq query --use_legacy_sql=false < data_lake/schemas/staging_tables.sql
# 4. Start the MCP server (trial mode by default — safe)
python mcp_server/server.py
# 5. Start the monitoring dashboard
streamlit run dashboard/daily_monitoring.py
# 6. Run tests
pytest tests/ -v
Core Workflow
1. ingest_carrier_statement(file, carrier, mode="trial")
↓ Parses PDF or Excel, normalizes with Claude LLM,
validates against BigQuery, stages for review
2. Daily: accounting team reviews dashboard at localhost:8501
↓ Reviews exception queue, approves/rejects transactions
3. When ready for live: set mode="live" per carrier
↓ Auto-posts ≥95% confidence transactions to Epic
↓ <95% still go to human review queue
Supported Carriers
- Nationwide (Excel)
- Travelers (PDF)
- (Add more — see docs/adding_carriers.md)
Documentation
| Doc | Contents |
|---|---|
CLAUDE.md |
Full project context for Claude Code sessions |
docs/adding_carriers.md |
How to onboard a new carrier |
docs/trial_run_guide.md |
Guide for the accounting team |
docs/epic_sdk_setup.md |
Applied Epic SDK configuration |
carrier_configs/carrier_template.yaml |
Template for new carrier configs |
Architecture
Carrier Files (PDF/Excel/Portal)
→ Ingestion (pdfplumber / pandas / playwright)
→ Normalization (Claude LLM → canonical schema)
→ Validation (BigQuery: policy match, duplicate check)
→ Confidence Scoring (auto ≥95%, review 80-94%, reject <80%)
→ Staging (BigQuery shadow table in trial / live table in live)
→ Applied Epic SDK (live mode only)
→ Audit Trail (BigQuery)
→ Daily Dashboard (Streamlit)
Safety Features
- Trial mode default: Zero Epic writes until explicitly switched to live
- Confidence thresholds: Only ≥95% confidence transactions auto-post
- Duplicate detection: Prevents double-posting
- Rollback support: Can void Epic entries if needed
- Full audit trail: Every transaction tracked source → Epic entry ID
- Human review queue: Exceptions always go to accounting team before posting
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.