Bridge Town MCP
Bridge Town exposes its financial planning capabilities through the Model Context Protocol (MCP). Connect your AI agent once — then ask Bridge Town questions directly from your conversation.
What is Bridge Town MCP?
Bridge Town is an MCP-native financial planning platform. Instead of building its own AI, Bridge Town provides a set of MCP tools that any compatible agent can use: create projects, write models, run them in a sandboxed environment, query results, and view version history.
Your agent stays in control. Bridge Town provides the storage, compute, version control, and audit trail.
MCP server endpoint
https://api.bridgetown.builders/mcpTransport: MCP Streamable HTTP. Claude.ai connects via OAuth 2.0 (no token needed). Other clients — Claude Code, Cursor, OpenCode — use a bearer token generated from app.bridgetown.builders/connect.
Connect your agent
Choose the client your team already uses:
| Client | Method |
|---|---|
| Claude.ai | 1-click connector |
| Claude Desktop | 1-click connector |
| Claude Code | claude mcp add command |
| Any MCP client | Manual JSON config |
Full step-by-step instructions are in the getting started docs.
What can your agent do?
Once connected, your agent can:
- Create and manage projects — version-controlled workspaces for financial models
- Write and run models — Python models executed in an isolated sandbox; no network access
- Query financial data — attach Google Sheets, CSV files, or uploaded data sources
- Compare scenarios — branch a model, change assumptions, see the financial delta before merging
- Generate dashboards — one-call HTML reports shareable as a link or file
- Review history — diff any two model versions; roll back to any prior state
For a complete list of available tools, see the MCP tool reference.
Scopes and access control
The MCP server uses token-scoped access. When you connect, you approve the scopes your agent needs. See MCP scopes for details on what each scope permits.
Security
Bridge Town never sends your financial data to a language model. Your agent runs on your infrastructure; Bridge Town is the storage and compute layer it reads from and writes to. Every tool call is logged in an append-only audit trail.
For the full security architecture, see the Security page.