Bridge Town Documentation

Build financial models by talking to your AI of choice. Bridge Town exposes its modelling, data, and execution tools over the Model Context Protocol — bring Claude, Codex, Gemini, or any MCP-compatible agent and it writes production Python into your versioned projects.

Get Started Quick start — sign up, connect your agent, and build your first model.

Agent Workflow Cookbook End-to-end worked example with MCP tool calls, responses, model code, dashboard creation, and sharing.

MCP Tools Reference Complete reference for every tool exposed via MCP.

Connect any MCP client Generic Streamable HTTP + bearer token instructions for Codex, OpenCode, ChatGPT-compatible, or custom agents.

Connect Claude Dedicated guides for Claude.ai (OAuth), Claude Code, and Claude Desktop.

Full MCP Tool Suite 75+ tools across modeling, data, dashboards, execution, Google Sheets, collaboration, and templates — all available via MCP.

Version History Every model change is versioned. Full project history, branching, and diff support.

Sandboxed Execution Models run in isolated Docker containers with --network none, read-only /repo + /data, and writable tmpfs at /outputs + /tmp.

Team Collaboration Invite teammates, review changes with pull requests, approve and merge, and publish reusable templates.

Multi-Model Pipelines Chain models together with the PIPELINE convention and branch-scoped /upstream transport.