Skip to main content
Glama
PyNet Bridge

PyNet Bridge

by RAEN-DT

PyNet Bridge is the execution layer that allows AI models to control desktop BIM tools in real-time.

It connects Natural Language → Python → desktop BIM tools (Navisworks, Revit, AutoCAD, Rhino, Tekla Structures, RIB Presto, AVEVA UE), enabling AI to generate, execute, and refine BIM workflows autonomously.

Available integrations include Navisworks Manage, Revit, Civil 3D, Rhino, Tekla Structures (2023–2026), RIB Presto (budgets, COM API) and AVEVA UE (E3D Design .NET API).

This bridge acts as the connective tissue between AI logic and the desktop APIs of those hosts, allowing for dynamic UI creation, script execution, and BIM process automation using natural language.

đŸŽŦ Demos & Tutorials

See PyNet Bridge in action — these recordings show the full natural-language → BIM-action workflow inside live Autodesk models.

Demo

Description

Video

⭐ Autonomous BIM Coordination (Revit + Navisworks)

Claude Code drives Revit and Navisworks simultaneously to audit models, resolve clashes by tolerance rules, generate geometry, and self-correct API errors in real time.

đŸŽŦ Watch here

PyNET + Codex Integration

How to configure PyNet Bridge with Codex and query into Navisworks.

đŸŽŦ Watch here

In this technical demo, Claude Code connects to Navisworks and Revit at the same time to run a full coordination workflow end-to-end:

  • Engineering criteria as a Skill — the AI applies tolerance rules to decide which interferences to auto-approve.

  • Real-time data auditing — validates model integrity and detects missing parameters (PYNET_Classification) before coordinating.

  • Autonomous geometry generation — generates architectural sleeves (pasatubos) directly in Revit, with precise spatial rotation to match pipe angles.

  • Self-healing code — hits a live Revit API data-type exception, analyzes the failure, rewrites the script on the fly, and completes the task.

Related MCP server: A2A MCP Server

🔄 How it works

  1. The user describes a task in natural language.

  2. The AI generates a Python script.

  3. PyNet Bridge validates and sends the script.

  4. The PyNet plugin executes it inside Autodesk.

  5. Results are returned back to the AI.

This is what turns AI from a chatbot into an execution engine for BIM.


🚀 What makes PyNet Bridge powerful

  • AI → Action: Turns AI-generated code into real actions inside Navisworks/Revit

  • Real-time Execution: Run scripts instantly without leaving the BIM environment

  • Dynamic UI Creation: Let AI create tools, buttons and workflows on the fly

  • Reliable Communication: Fast and stable, entirely local

  • Model-Aware Automation: Operates directly on live BIM models


đŸ› ī¸ Installation

Open PowerShell and run:

irm https://raw.githubusercontent.com/RAEN-DT/PyNetBridge/main/install.ps1 | iex

This will automatically:

  1. Check Python 3.10+ is installed

  2. Install uv if missing, then pynet-mcp-bridge from PyPI with uv (never pip)

  3. Auto-detect and configure all installed AI clients:

    • Claude Desktop (standard and Microsoft Store versions)

    • Claude Code (VS Code extension / CLI)

    • Cline (VS Code extension)

    • Roo Code (VS Code extension)

    • Codex CLI (~/.codex/config.toml)

The pynet-mcp-bridge package includes:

Package

Purpose

pynet-mcp-bridge

MCP server that connects AI models with desktop BIM tools via PyNET

mcp[cli] (1.x)

Model Context Protocol SDK and CLI tools

psutil

System process detection (finds running Autodesk instances)

Restart your AI client(s) after installation to apply changes.

đŸ“Ļ Python Libraries Starter Pack (optional)

Install the recommended Python libraries for PyNET scripting:

irm https://raw.githubusercontent.com/RAEN-DT/PyNetBridge/main/install-libraries.ps1 | iex

This installs:

Library

Purpose

pandas

Data analysis and manipulation

plotly

Interactive charts and visualizations

matplotlib

Static plots and graphs

dash

Web dashboards from Python

These are the third-party libraries listed under Allowed Python Imports. Standard library modules (json, sys, re, etc.) are already included with Python.

Prerequisites

  • PyNet Platform plugin installed in Navisworks/Revit.

  • Python 3.10 or higher → python.org — Python 3.14 is supported.

  • uv → docs.astral.sh/uv — required. The Claude Desktop extension (.mcpb) launches the server with uvx pynet-mcp-bridge, so uv must be installed and on your PATH. Install it with winget install astral-sh.uv. Without uv, the extension will fail to start.

  • Git → git-scm.com — required for VS Code extensions (Claude Code, Cline, Roo Code) to function correctly.

  • For Cline / Roo Code: VS Code → code.visualstudio.com


🔧 Option B — Manual installation

1. Install the package:

uv tool install pynet-mcp-bridge

Use uv only — do not install with pip. A pip copy lands in a different interpreter from the pynet-bridge launcher, so upgrades and fixes silently miss the environment that actually runs.

2. Configure Claude Desktop:

Add the following to your claude_desktop_config.json:

  • Standard: %APPDATA%\Claude\claude_desktop_config.json

  • Microsoft Store: %LOCALAPPDATA%\Packages\Claude_*\LocalCache\Roaming\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "pynet-bridge": {
      "command": "pynet-bridge",
      "args": []
    }
  }
}

3. Configure Claude Code (VS Code extension):

Add to %USERPROFILE%\.claude.json:

{
  "mcpServers": {
    "pynet-bridge": {
      "type": "stdio",
      "command": "pynet-bridge",
      "args": []
    }
  }
}

4. Configure Cline:

Add to %APPDATA%\Code\User\globalStorage\saoudrizwan.claude-dev\settings\cline_mcp_settings.json:

{
  "mcpServers": {
    "pynet-bridge": {
      "type": "stdio",
      "command": "pynet-bridge",
      "args": []
    }
  }
}

5. Configure Roo Code:

Add to %APPDATA%\Code\User\globalStorage\rooveterinaryinc.roo-cline\settings\mcp_settings.json:

{
  "mcpServers": {
    "pynet-bridge": {
      "type": "stdio",
      "command": "pynet-bridge",
      "args": []
    }
  }
}

6. Configure Codex CLI:

Add to %USERPROFILE%\.codex\config.toml:

[mcp_servers.pynet-bridge]
command = "C:/Users/<user>/.local/bin/pynet-bridge.exe"
args = []

đŸ› ī¸ Available MCP Tools

These tools allow AI to fully control the PyNet environment, from UI creation to script execution and system monitoring. Once connected, the AI will have access to the full suite of PyNet tools:

🧠 Core capabilities exposed to AI

🔍 System & Connection

  • list_active_instances: Scans the system for running Autodesk processes with an active PyNet connection.

  • check_plugin_status: Handshake ping to verify the plugin listener is responsive.

đŸ—ī¸ Module (Tab) Management

  • get_pynet_ui_layout: Fetches the full UI structure (ButtonsModules and ScriptButtons).

  • create_pynet_module: Creates a new custom Tab (ButtonsModule) in the Ribbon.

  • delete_pynet_module: Permanently deletes a module and all its contents.

🔘 Button Management

  • get_buttons_data: Lists all script buttons for a specific module ID.

  • deploy_script_button: Installs a new ScriptButton into a specific module (Name, Script, Icon, Tooltip).

  • update_script_button: Updates metadata for an existing ScriptButton or moves it to another module.

  • delete_script_button: Permanently removes a ScriptButton from a module by Id.

đŸ’ģ Execution & Console Control

  • send_command: Direct script execution in the PyNet engine (Target PID, Script Name, Content).

  • get_output_window_status: Checks if the output window is currently available/visible.

  • configure_output_window: Toggles the visibility of the PyNet log/output window.

🧊 BIM Viewer Control (VS Code)

Drives the PyNet BIM Viewer open in VS Code. Reads clash data from the loaded .pnt package and controls the 3D view over a local (127.0.0.1) connection — nothing leaves your machine.

  • viewer_status: Reports whether a PyNet BIM Viewer is open (port, package, data dir).

  • viewer_list_clashes: Reads the loaded package's clashes.json and returns each clash with its pnt_id identifiers.

  • viewer_load_package: Loads a .pnt into the open viewer, pushes the models, and returns a summary.

  • viewer_highlight_clash: Highlights a clash pair by pnt_id (element A red, element B green).

  • viewer_fit: Fits the camera to all models in the viewer.

  • viewer_clear: Clears highlights and ghosting in the viewer.


💡 Usage Examples

These are realistic prompts you can give your AI client once PyNet Bridge is connected.

Example 1 — Discover a running model and query it (read-only)

Prompt: "Connect to my open Navisworks model and tell me how many items are in the current selection."

The AI calls list_active_instances to find the running Navisworks process (e.g. Navisworks (PID 12345)), then calls send_command with a short Python script that reads the active document's selection. Expected output: a message like Current selection contains 42 items.

Example 2 — Run a clash-detection summary

Prompt: "Give me a summary of clash test results in the open model and export the counts as a table."

The AI calls list_active_instances, then send_command with a script that iterates the DocumentClash results using the whitelisted Autodesk.Navisworks.Clash assembly and returns counts per test. Expected output: a table such as Test 'Structure vs MEP': 17 active, 3 resolved.

Example 3 — Build a custom ribbon tool on the fly

Prompt: "Create a new ribbon tab called 'QA' and add a button that runs my hide-empty-layers script."

The AI calls create_pynet_module (returns a new module ID), then deploy_script_button with the button name, script path, icon and tooltip. Expected output: Button 'Hide Empty Layers' deployed to module QA. — visible immediately in the Autodesk ribbon.

â„šī¸ Every script sent via send_command / send_command_by_path is first validated by the built-in static analyzer (see Safe AI Execution). Scripts that violate the import/call whitelist are rejected before reaching Autodesk.


đŸ›Ąī¸ Safe AI Execution

PyNet Bridge includes a built-in validation layer that ensures all AI-generated scripts are safe and controlled before execution.

✔ Prevents unsafe operations
✔ Blocks unauthorized system access
✔ Guarantees controlled interaction with BIM models

AI remains powerful, but within safe boundaries

Starting from v1.1.1, the MCP server includes a built-in static analyzer that validates every script before it reaches the host application. All scripts are parsed and inspected at the bridge level — rejected scripts never leave the MCP server.

Allowed CLR Assemblies

Only these .NET references are permitted via clr.AddReference:

  • Common: System, System.Windows.Forms, System.Drawing, System.Collections.Generic

  • Navisworks: Autodesk.Navisworks.Api, .ComApi, .Interop.ComApi, .Clash

  • Revit: RevitAPI, RevitAPIUI

  • AutoCAD / Civil 3D / Map 3D: AcMgd, AcCoreMgd, AcDbMgd, AecBaseMgd, AecPropDataMgd, AeccDbMgd, ManagedMapApi

  • Rhino: RhinoCommon, Rhino.UI, Eto

  • Tekla Structures: Tekla.Structures, .Model, .Drawing, .Datatype, .Dialog, .Catalogs, .Plugins

  • RIB Presto: none needed — scripts drive Presto through the __presto__ scope variable (COM)

  • AVEVA UE (E3D Design): Aveva.Core.Database, .Geometry, .Utilities, Aveva.ApplicationFramework, .Presentation, PMLNet

  • PyNet plugins: Raen.Core.Pynet.*, Raen.{Product}.Pynet.* (any version — e.g. Raen.Core.Pynet.Resources, Raen.Navisworks.Pynet.2024, Raen.Civil3D.Pynet.2026, Raen.Rhino.Pynet.8, Raen.Tekla.Pynet.2026, Raen.Presto.Pynet, Raen.Aveva.Pynet.*)

Allowed Python Imports

clr, sys, json, re, time, datetime, pathlib, typing, threading, collections, xml, math, pandas, plotly, matplotlib, dash, webbrowser, psutil, functools, openpyxl, uuid, zipfile, io, mimetypes, difflib, csv, ifcopenshell, numpy, shapely, qgis, processing, pypdf, docx, rhinoscriptsyntax, scriptcontext

Allowed Python Submodules

Some modules are allowed at the submodule level only, preventing access to dangerous siblings:

Allowed

Blocked

Reason

http.server

http.client, http.cookiejar

Allow local HTTP serving, block outbound requests

Blocked Python Imports

os, subprocess, shutil, socket, ctypes, pickle, importlib, urllib, signal, multiprocessing, tempfile, glob, inspect, code, codeop

Blocked Calls

eval, exec, compile, __import__, getattr, setattr, delattr, globals, locals, vars, breakpoint, open

Blocked Attribute Access

__builtins__, __subclasses__, __globals__, __code__

Any script that violates these rules is immediately rejected with a descriptive error message, without ever being sent to the plugin.


📂 Project Structure

  • pynet_mcp/: Core MCP server logic (FastMCP).

  • pyproject.toml: Package configuration and dependency management.


đŸ“Ĩ Getting Started

Start building autonomous BIM workflows in minutes.

Install the bridge, connect your AI client, and turn natural language into real actions inside your models.


❓ FAQs

Have questions about installation, configuration, or usage? Check the full FAQ page:

👉 PyNet FAQs


🔗 How This MCP Fits Into the Ecosystem

This MCP is part of a modular system designed to enable AI-driven BIM automation across desktop BIM tools.

This repository is designed to work alongside:

  • PyNet Platform → Executes scripts inside Navisworks, Revit, Civil 3D, Rhino, Tekla Structures, RIB Presto & AVEVA UE via Python.NET

  • PyNet Library → Gives the AI context with a Python scripts library

Together, these components enable:

Natural Language → AI → Python Script → PyNet → Host application → BIM Action

Component

Repository

Purpose

PyNet Platform

PyNet

Navisworks, Revit, Civil 3D, Rhino, Tekla Structures, RIB Presto & AVEVA UE plugin — hosts the Python.NET engine

PyNet Bridge (MCP)

This repo

MCP server - connects AI models to PyNET with including secure scripts validation

PyNet Library

PyNetLibrary

Script reference library and AI context

🔒 Privacy Policy

PyNet Bridge runs entirely on your local machine. It does not collect, store, or transmit any personal data, telemetry, or analytics.

  • No data collection: The server does not send any information to RAEN Digital Tools or any third party.

  • Local-only communication: All communication between the MCP server and the PyNet plugin running inside Autodesk stays on your machine. No network sockets, no remote endpoints — nothing leaves your computer.

  • Your scripts and model data stay on your machine and are only processed by the AI client you have connected.

Full privacy policy: https://privacy.raendt.com/

🆘 Support

đŸ–Ĩī¸ Platform Support

PyNet Bridge is Windows-only. It relies on Windows-specific local communication facilities and on desktop BIM applications (Navisworks, Revit, Civil 3D, Rhino, Tekla Structures, RIB Presto, AVEVA UE), which are Windows products. macOS and Linux are not supported.

📄 License

This project is licensed under the MIT License.


mcp-name: io.github.RAEN-DT/pynet-mcp-bridge

Available Tools

24 tools
check_plugin_statusB
Read-only

Handshake ping to verify the plugin listener is responsive.

Inputs: pid (required). Returns: result

configure_output_windowC
Destructive

Toggles the visibility of the PyNet log/output window.

Inputs: pid (required), is_available (required). Returns: result

create_pynet_moduleC
Destructive

Creates a new custom Tab (ButtonsModule) in the Ribbon.

Inputs: pid (required), name (required). Returns: result

delete_pynet_moduleC
Destructive

Permanently deletes a module and all its contents.

Inputs: pid (required), module_id (required). Returns: result

delete_script_buttonB
Destructive

Permanently removes a ScriptButton from a module by Id.

Inputs: pid (required), button_id (required), module_id (required). Returns: result

deploy_script_buttonB
Destructive

Installs a new ScriptButton into a specific module (Name, Script, Icon, Tooltip).

Inputs: pid (required), name (required), module_id (required), script_Path (required), +2 more. Returns: result

get_buttons_dataA
Read-only

Lists all script buttons for a specific module ID.

Inputs: pid (required), module_id (required). Returns: result

get_output_window_statusB
Read-only

Checks if the output window is currently available/visible.

Inputs: pid (required). Returns: result

get_pynet_ui_layoutC
Read-only

Fetches the full UI structure (ButtonsModules and ScriptButtons).

Inputs: pid (required). Returns: result

list_active_instancesA
Read-only

Scans the system for running host processes (Autodesk, Rhino, AVEVA UE, PyNET for Tekla/Presto) with an active PyNet IPC pipe.

Inputs: no named parameters. Returns: result

send_commandB
Destructive

Direct script execution in the PyNet engine (Target PID, Script Name, Content).

Inputs: pid (required), content (required), timeout (required), script_name (required). Returns: result

send_command_by_pathA
Destructive

Executes a script file directly by path in the PyNet engine, without sending content inline.

Inputs: pid (required), timeout (required), file_path (required), script_name (required). Returns: result

update_script_buttonB
Destructive

Updates metadata for an existing ScriptButton or moves it to another module.

Inputs: pid (required), name (required), tooltip (required), button_id (required), +4 more. Returns: result

viewer_clearA
Read-only

Clears highlights and ghosting in the open viewer.

Inputs: no named parameters. Returns: result

viewer_fitA
Read-only

Fits the camera to all models in the open viewer.

Inputs: no named parameters. Returns: result

viewer_get_propertiesA
Read-only

Reads element properties (name, model, psets) from the loaded package's properties.json.

Pass one or more pnt_ids to get their full properties (including psets) directly.

Called with no pnt_ids, returns a paginated lightweight index (pnt_id, name, model) instead of every element — narrow it with model (exact model/discipline name, e.g. "Snowdon Towers Sample HVAC") and/or search (case-insensitive substring of the element name), and page through with limit/offset (default 200 per page, capped at 500). On a federated model the unfiltered index can have tens of thousands of entries — always pass model or search when you can, and use offset to page through the rest.

The data source is the .pnt's properties.json, NOT the viewer or IFC.

Inputs: limit, model, offset, search, +1 more. Returns: result

viewer_get_stateA
Read-only

Reads the viewer's last reported state: loaded models and the pnt_id of whatever the user currently has selected (click in the 3D view or the tree panel — null if nothing is selected). Use the returned pnt_id with viewer_get_properties for its details, or feed it into a clashes.json lookup to find what it clashes with.

Inputs: no named parameters. Returns: result

viewer_highlight_clashA
Destructive

Highlights a clash (element(s) A red, element(s) B green) and isolates the rest away.

Usually one pnt_id per side (a single clash pair), but either side accepts a list — e.g. one element (A) against every counterpart it clashes with (B), all shown at once.

Inputs: pnt_id_a, pnt_id_b. Returns: result

viewer_isolateA
Destructive

Isolates (hides everything except) the given pnt_ids in the open viewer.

Inputs: pnt_ids. Returns: result

viewer_isolate_modelsA
Destructive

Shows only the given model(s)/discipline(s) (by exact name, e.g. "Snowdon Towers Sample HVAC"), hiding every other loaded model. Cheaper than viewer_isolate for "just show me this discipline" requests — no need to resolve every element's pnt_id first. Get the exact model names from viewer_get_state's "models" list. Undo with viewer_clear.

Inputs: models. Returns: result

viewer_list_clashesA
Read-only

Reads the loaded package's clashes.json — the data source, NOT the viewer/IFC.

Returns each clash with the pnt_id identifiers needed to highlight it.

Inputs: no named parameters. Returns: result

viewer_load_packageB
Destructive

Loads a .pnt into the open viewer and returns a summary read from clashes.json.

Inputs: pnt_path (required). Returns: result

viewer_selectA
Destructive

Highlights a set of elements in the open viewer by pnt_id, in neutral colours (yellow / blue) with nothing else hidden — a plain "look at these" pointer.

Pass a list of pnt_ids in pnt_ids (group A). Optionally pass a second list in pnt_ids_b to highlight a second group in a distinct colour. Replaces any previous selection. Use viewer_list_clashes / viewer_get_properties to obtain pnt_ids. For a clash pair (red/green, with the rest of the model isolated away), use viewer_highlight_clash instead.

Inputs: pnt_ids, pnt_ids_b. Returns: result

viewer_statusA
Read-only

Reports whether a PyNet BIM Viewer is open in VS Code (port, package, data dir).

Inputs: no named parameters. Returns: result

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelogâ€Ļ

    TDQS

    B3.3/5.0

    Scored across 24 tools

    Disambiguation4/5

    The viewer tools have the most potential for confusion (select vs highlight_clash vs isolate vs isolate_models vs clear), but the descriptions explicitly cross-reference each other and state when to prefer one over another, which resolves most of it. A few boundaries remain soft — check_plugin_status vs list_active_instances vs viewer_status all report 'is something alive', and configure_output_window vs get_output_window_status are near-mirror pairs — but they are separable.

    Naming Consistency4/5

    Nearly everything follows verb_noun (get_, create_, delete_, update_, deploy_, list_, send_, configure_, check_), with a clean viewer_ namespace for the 3D side. Minor deviations: the ribbon/button noun is inconsistent ('buttons' in get_buttons_data vs 'script_button' in deploy/update/delete_script_button, and get_pynet_ui_layout is a noun-first outlier), and some tools carry a domain prefix while others do not.

    Tool Count3/5

    24 tools is on the heavy side, and the surface spans four separate concerns (viewer control, ribbon/module management, IPC process discovery, script execution). The count is defensible given the breadth of the PyNet bridge, but a few low-value entries (viewer_fit, get_output_window_status) could be folded into neighbours.

    Completeness4/5

    Script buttons have full CRUD (deploy/update/delete/get_buttons_data), the viewer has load/state/properties/select/isolate/highlight/clear/fit, and execution has both inline and by-path variants. The main gap is module lifecycle: create_pynet_module and delete_pynet_module exist but there is no update/rename/reorder module tool, and there is no way to unload a package or detach from a running instance.

    Maintenance

    ActivityActive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers