What I Built

I built Sybil: Real-Time Risk Intelligence, an autonomous sentinel designed to bridge the gap between operational execution and strategic decision-making. In any high-stakes organization, silos are the biggest threat; a delay in one department can trigger a chain reaction of losses in another if it is not communicated instantly.

Sybil acts as a 24/7 intelligence layer. It proactively monitors Notion databases to detect conflicts, calculates the potential impact of delays, and automatically populates a dedicated Alert Center. By turning static data into live alerts, Sybil ensures that stakeholders are never blindsided by uncommunicated changes.

Show us the code

Erika05cristina / Sybil-mcp

🧠 Sybil MCP — Notion Risk Intelligence Server

Sybil is an MCP (Model Context Protocol) server that monitors your Notion databases in real time and detects financial risks: when an Engineering task is delayed and overlaps with a Marketing campaign launch, Sybil automatically creates an alert in your Notion Alert Center.

⚡ Quick Start

You don't need to run anything manually. Claude Desktop manages the server automatically.

Make sure your claude_desktop_config.json is configured (see Configuration section)
Open Claude Desktop — the server starts on its own in the background
Ask Claude: "What risks does Sybil detect?"

⚠️ Never run node index.js manually while Claude Desktop is open. It would create two server instances running in parallel and duplicate alerts in Notion.

🏗️ Architecture

┌─────────────────────────────────────────────────────────────┐
│                      Claude Desktop                         │
│                                                             │
│   ┌─────────────┐   MCP Protocol   ┌────────────────────┐  │
│   │  Claude AI  │ ◄──────────────► │  Sybil MCP Server  │  │
│   │

…

View on GitHub

How I Used Notion MCP

I integrated Notion MCP to unlock a truly proactive AI assistant that does not wait for user input to protect the business:

1. Cross-Database Intelligence: Using the Model Context Protocol, Sybil audits multiple Notion databases simultaneously every 10 seconds. This allows the AI to find hidden correlations between disparate data sets that humans might miss.
2. Native Tool Integration: I registered the check_sybil_risk tool, allowing Claude Desktop to perform deep-dive audits on demand. This transforms the AI from a simple chatbot into a specialized Risk Consultant with direct access to the "source of truth."
3. Automated Intervention: MCP enables Sybil to act as a bridge. When a risk is identified, the server performs a multi-step intervention: writing context-aware comments on the original task and creating high-priority entries in the Alert Center.
4. Smart Session Management: I implemented a robust anti-spam engine using local memory and Notion-side validation. This ensures that even with a persistent 10-second monitoring cycle, the system remains efficient and the data stays clean.