Bridge Town

official

Bridge Town is an MCP-native, git-versioned financial modeling platform for FP&A teams and finance leaders. AI agents use Bridge Town tools to create projects, write Python model files, run models in isolated cloud sandboxes, query data, write outputs to Google Sheets, create dashboards, branch scenarios, and collaborate with teammates.

GitHub
1
Try this MCPSponsored

What can you do with Bridge Town MCP?

  • Discover the right tool — Ask your assistant to call search_tools to find the correct current MCP tool for a financial modeling goal.
  • Inspect tool schemas — Use get_tool to retrieve the full schema of any tool before invoking it.
  • Preserve sheet states — Run native_sheet_reader.py in a model repo to read Native Sheets while keeping missing, blank, and formula states intact.
  • Track output dependencies — Generate validated output_lineage.json dependency graphs with output_lineage.py for model outputs.
  • Connect via stdio — Forward MCP calls to the hosted endpoint using the local stdio shim with a Bridge Town API token.

Documentation

Bridge Town MCP Connector

Build financial models as code with cloud execution, Native Sheets, Google Sheets, version control, and collaboration.

This public repository contains documentation, registry metadata, a minimal stdio-to-HTTP shim, and versioned model-authoring helpers for the hosted Bridge Town MCP connector. The commercial server implementation is operated by Bridge Town and is not published here.

Bridge Town does not run a server-side LLM or proxy prompts. The user's chosen AI client and model provider run the conversation; Bridge Town receives typed MCP calls and executes deterministic platform operations.

Connect

For Claude.ai, add https://api.bridgetown.builders/mcp as a custom connector and approve the Bridge Town OAuth prompt.

For clients that support Streamable HTTP plus headers:

{
  "mcpServers": {
    "bridge-town": {
      "transport": {
        "type": "http",
        "url": "https://api.bridgetown.builders/mcp",
        "headers": {
          "Authorization": "Bearer btk_YOUR_TOKEN"
        }
      }
    }
  }
}

Generate and revoke API tokens at https://app.bridgetown.builders/connect. See Getting Started for setup, current task flows, and troubleshooting.

Canonical MCP surface

The production endpoint exposes one canonical tool catalog. Use search_tools to find the correct current tool for a goal and get_tool to retrieve its full schema. The catalog does not change based on the client or user agent.

The canonical registry manifest for this public listing is server.json. It points to the hosted remote MCP server; it is not a copy of the proprietary server implementation.

Versioned model-authoring helpers

Small helper modules that are intended to run inside a model repository are published as immutable, MIT-licensed releases:

The v1.1.0 release directory includes runnable examples, a changelog, ownership, license, manifest, and SHA256SUMS. Install a helper byte-for-byte at the manifest's target_path and verify its digest before committing it to a model repository.

Local stdio shim

Most clients should connect directly to the Streamable HTTP endpoint. A local MCP host that requires stdio may use the minimal forwarding shim:

BRIDGE_TOWN_API_TOKEN=btk_YOUR_TOKEN npx github:Bridge-Town/financial-modeling-mcp

The shim does not read or write local files or spawn child processes. It only forwards MCP JSON-RPC messages over HTTPS to the configured Bridge Town MCP endpoint.

Security

Financial model execution runs in an isolated container without outbound network access. Google access uses drive.file, limiting Bridge Town to files selected through the Bridge Town flow or created by Bridge Town.

Report vulnerabilities privately through SECURITY.md. Do not open a public issue or include customer data in a report.

Governance

The governance files are generated from the closed, versioned directory-resources.v1.json metadata contract.