MCP Connector for Navisworks Documentation
Complete guide to installing, configuring, and using MCP Connector for Navisworks - 74 MCP tools, including the 25 free read-only tools, driven by Claude Desktop, ChatGPT Desktop, Cursor, or Claude Code.
Installation
Get up and running in minutes
Configuration
Configure Claude, ChatGPT & Cursor
Tool Reference
74 tools with example prompts
MCP Connector for Navisworks connects AI assistants to a live Autodesk Navisworks session on your machine. Once connected, your AI can read and author the drawing: query entities, create and modify geometry, manage layers, blocks, external references, annotation, layouts and 3D solids. Two parts ship together and talk over a local connection: an MCP server your AI client launches, and a plug-in that loads inside Navisworks. Nothing needs to be configured between them.
Supported Navisworks Versions
- Navisworks 2022
- Navisworks 2023
- Navisworks 2024
- Navisworks 2025
- Navisworks 2026
- Navisworks 2027
Full Navisworks is required. Navisworks LT is not supported. LT does not run add-ins, and the connector works as a plug-in inside Navisworks. Windows 10 or 11, 64-bit.
AI Applications
- Claude Desktop
- ChatGPT Desktop
- Cursor
- Claude Code
Installation
Before you begin
Make sure all running instances of Autodesk Navisworks are closed before starting the installation.
1
Download the Installer
Download the latest version of MCP Connector for Navisworks from the product page.
2
Run the Installer
Double-click the downloaded
.exe file and follow the installation wizard, accepting the licence agreement and privacy policy. One install covers every supported Navisworks release, so there is nothing to choose.3
Launch Navisworks
Open Autodesk Navisworks. Look for the AUTOM8LABS tab on the ribbon. The plug-in auto-loads on startup, so the tab appears on its own; if it is missing, check the startup log noted under Troubleshooting.
4
Start Using MCP Connector for Navisworks
Click the MCP Connector button to open the connection dialog and auto-configure your AI application.
Enabling MCP Connector for Navisworks
After installation, click the MCP Connector button on the AUTOM8LABS tab to open the connector dialog, then toggle Enable MCP to start the connection. No licence key is needed for the free tools - the key only unlocks the Pro set.
The dialog shows the status of detected AI applications:
- Ready (Running) - The AI application is configured to reach Navisworks and is currently open
- Ready - Configured, but the application is not running right now
- Detected - The application is installed but not yet configured; toggle Enable MCP to configure it
- Not Detected - The AI application is not installed on this machine
The indicator at the top of the dialog reads Disabled, Waiting, or Active. Toggling Enable MCP off again removes the connector from every detected client's configuration.
Configuration
MCP Connector for Navisworks includes one-click auto-configuration for supported AI applications. Toggling Enable MCP in the connector dialog writes the configuration for every detected application at once. The installer also registers the server on install, so in most cases there is nothing to do by hand.
Claude Desktop
The config file is located at:
Path
%APPDATA%\Claude\claude_desktop_config.json
Using the Microsoft Store version of Claude Desktop?
The Microsoft Store build never reads %APPDATA%\Claude - Windows redirects that folder into a sandboxed package location. MCP Connector for Navisworks detects the Store package and writes to both paths, so the Store build is configured automatically. The redirected file lives here:
%LOCALAPPDATA%\Packages\Claude_<hash>\LocalCache\Roaming\Claude\claude_desktop_config.json
If tools still do not appear, fully quit Claude from the system tray (closing the window is not enough) and reopen it.
ChatGPT Desktop
ChatGPT appears in the dialog when it is installed on your machine. The config file is located at:
Path
%USERPROFILE%\.codex\config.toml
This file is shared with the Codex CLI, so configuring ChatGPT Desktop makes the Navisworks tools available in Codex CLI as well. After configuring, fully quit ChatGPT and reopen it. With Navisworks open and a drawing loaded, a quick test is to ask ChatGPT to "ping Navisworks".
Use Work, not Chat. ChatGPT Desktop loads MCP servers from this config file in its Work area only - regular Chat conversations cannot see the Navisworks tools. If ChatGPT replies that the Navisworks MCP is not connected, check you are in Work first, then start a fresh conversation - the MCP connection is per-conversation and drops after sitting idle.
Cursor
The config file is located at:
Path
%USERPROFILE%\.cursor\mcp.json
Claude Code & other MCP clients
MCP Connector for Navisworks exposes a standard MCP stdio server, so it also works with Claude Code, VS Code, and any other MCP-compatible client. The server is a single executable installed with the plug-in bundle:
Path
C:\ProgramData\AUTOM8LABS\MCPConnectorNavisworks\Bridge\MCPConnectorNavisworks.Bridge.exe
Register it in Claude Code from a terminal:
Command Prompt
claude mcp add AUTOM8LABS_Navisworks -- "C:\ProgramData\AUTOM8LABS\MCPConnectorNavisworks\Bridge\MCPConnectorNavisworks.Bridge.exe"
Manual configuration (any MCP client)
Add this to your client's MCP server configuration:
JSON
{
"mcpServers": {
"AUTOM8LABS_Navisworks": {
"command": "C:\\ProgramData\\AUTOM8LABS\\MCPConnectorNavisworks\\Bridge\\MCPConnectorNavisworks.Bridge.exe"
}
}
}VS Code
VS Code uses a slightly different shape. Add the server to your MCP configuration (for example .vscode/mcp.json in a workspace):
JSON
{
"servers": {
"AUTOM8LABS_Navisworks": {
"type": "stdio",
"command": "C:\\ProgramData\\AUTOM8LABS\\MCPConnectorNavisworks\\Bridge\\MCPConnectorNavisworks.Bridge.exe"
}
}
}Restart Required
After changing MCP configuration, restart the AI client completely for the changes to take effect.
Claude Desktop Settings
MCP Connector for Navisworks exposes 74 tools. In Claude Desktop, enable Load tools when needed so tools are loaded on demand rather than all at once.
Use Chat, not Cowork
Claude Desktop has a mode selector at the bottom of the message box: Chat and Cowork. MCP connectors run in Chat conversations. In Cowork, your Navisworks requests will not reach the connector and Claude behaves as if Navisworks is not connected - even though the connector is configured correctly. Before you start, check the selector is set to Chat.
Note this is the opposite way round to ChatGPT Desktop, where the Navisworks tools are available in Work rather than regular Chat.
Operator Skill
MCP Connector for Navisworks ships with a free operator skill - a plain text playbook Claude reads before touching your drawing. It teaches Claude to use the connector efficiently: read in bulk instead of one entity at a time, verify answers with counts, batch edits in one call, and say so instead of guessing when something cannot be done. The result is fewer calls, fewer dead chats, and far less of your usage allowance burned on a single question.
Install in Claude Desktop:
- Download the skill: navisworks-connector.zip
- In Claude Desktop, go to Settings > Skills > Add > Upload a skill and select the zip.
- That is it - Claude picks the skill up automatically in new chats. If you ever find it has not, mention it in your prompt, for example "Follow the navisworks-connector skill".
Using Claude Code or Cursor instead? Install steps for both are in the mcp-connector-skills repository, which also hosts free skills for the rest of the MCP Connector family.
Free vs Pro
MCP Connector for Navisworks adds 74 MCP tools - 25 free and 49 Pro. Two editions ship in one install:
- Free is read-only inspection: a connectivity check, plus tools that query and inspect the federated model, walk the selection tree, read item properties and grids, review clash results and 4D tasks, measure, and navigate the view with zoom. Nothing in the free edition creates, modifies, or exports. No key is needed; it just works.
- Pro unlocks everything else: the full clash workflow (create and run tests, tune ignore rules, group, status, assign and comment on results), BCF 2.1 issue export with viewpoints and snapshots, sectioning, selection and search sets, appearance overrides, saved viewpoints and camera fly-throughs, 4D construction sequencing, quantity export to CSV, model appending and NWD publishing.
Pro is a per-seat subscription: £10 per seat per month or £100 per seat per year. See the product page for purchasing.
In the tool reference below, every tool is marked Free or Pro. On a free install, Pro tools are not listed to the AI client at all, and the Navisworks plug-in refuses to execute them, so there is no way to run a paid tool without a valid licence.
Edition changes need a client restart
The MCP server reads the edition when your AI client starts it. After activating or deactivating a licence, restart (or reconnect) the AI client to refresh the tool list.
Licence Activation and Management
Licensing lives entirely in the Navisworks plug-in. There is no key to set in the AI client.
1
Open the licence dialog
In Navisworks, click MCP Connector on the AUTOM8LABS ribbon panel, then Activate Key in the dialog footer. The same dialog is reachable from the Manage Licence command under the Tool add-ins tab. This opens the same dialog directly.
2
Enter your key and activate
Paste the licence key from your purchase email and click Activate. The plug-in validates the key with the AUTOM8LABS licence server and activates this machine's seat.
3
Reconnect your AI client
Restart your AI client so it picks up the full Pro tool list.
- One seat per machine. A seat covers every AI client on that machine: Claude Desktop, Claude Code, Cursor, VS Code, and any other MCP client.
- Moving machines. Click Deactivate in the licence dialog to free the seat, then activate on the new PC.
- Working offline. A validated key is honoured offline for 14 days, so short offline stretches never block your work.
Quickstart
1
Open Navisworks and a drawing
The plug-in auto-loads on startup. The AUTOM8LABS tab appearing on the ribbon confirms it is running.
2
Check the connection
In your AI client, ask it to run the
ping tool. You should get back "Connected to Navisworks."3
Ask your AI about the drawing
Try: "List the layers in this drawing and tell me which ones have nothing on them." The AI reads the layer table and entity counts and reports back. On Pro, follow up with "Freeze the empty ones."
Tool Reference
MCP Connector for Navisworks adds 74 MCP tools, grouped by category below, including the 25 free read-only tools. Every tool name shown is the exact identifier exposed to your AI client. The Edition column shows whether a tool is available in the free edition or needs a Pro licence. Under each description is an example prompt showing how to ask for it in natural language - click any prompt to copy it, paste it into your AI client, and adjust the names and numbers to suit your model.
Connectivity
| Tool | Example prompt | Edition |
|---|---|---|
| Check that the AI client can reach a running Navisworks session | ||
ping | Ping Navisworks and tell me if you are connected | Free |
Model and Document
| Tool | Example prompt | Edition |
|---|---|---|
| Get an overview of the open Navisworks document: title, file name, units, number of aggregated models, and overall bounding box. | ||
get_model_info | What model is open in Navisworks and what are its units? | Free |
| Get the display units of the open document (e.g. Meters, Millimeters, Feet). | ||
get_units | What units is this model in? | Free |
| List the grid systems in the scene and their levels (name + elevation). Useful for locating clashes or elements by grid line and storey. | ||
list_grids | List the grid lines and levels in this model | Free |
| List the aggregated source models in the scene — one entry per appended file, with its source file name, units and creating application. | ||
list_models | List every model appended into this federation | Free |
| Save the open model file in place, back to the path it was opened from. For a model that has never been saved, write it out with export_nwd or export_nwf instead. | ||
save_document | Save the current Navisworks file | Pro |
Properties and Query
| Tool | Example prompt | Edition |
|---|---|---|
| Search the whole scene for items by property values. Give one criterion via category/property/value, or several via 'conditions' (all must match). Category and property names as shown in the Properties window (e.g. category \, property \). Returns matching items with their ids. | ||
find_items | Find every item where Category is Ducts on Level 3 | Free |
| Get the combined bounding box (min/max/center/size) of one or more items addressed by their ids. | ||
get_bounding_box | What is the bounding box of the current selection? | Free |
| List an item's ancestry in the model tree, from the model root down to the item itself (root-first), plus the source file that owns it. Useful for locating an item in the tree. | ||
get_item_ancestors | Show the model tree path for the selected item | Free |
| Read all property categories and properties of a single model item, addressed by its id (from get_selection_tree, get_current_selection, or find_items). | ||
get_item_properties | Show me all properties of the selected item | Free |
Selection
| Tool | Example prompt | Edition |
|---|---|---|
| Clear the current selection so no items are selected. | ||
clear_selection | Clear the current selection | Pro |
| List the items currently selected in Navisworks, with their ids so other tools can act on them. | ||
get_current_selection | What is currently selected? | Free |
| Get the number of items currently selected in Navisworks. | ||
get_selection_count | How many items are selected? | Free |
| Walk the model selection tree. With no id, returns the top-level model roots; with an item id, returns that item's children. 'depth' expands nested children (1-5). Each node includes an id for drilling deeper or reading properties. | ||
get_selection_tree | Show me the selection tree down two levels | Free |
| Search the whole scene for items whose property matches a value, then select those items. Give the property's category and name as shown in the Properties window (e.g. category \, property \). | ||
select_by_search | Select every pipe larger than 150mm on Level 2 | Pro |
| Activate a saved selection set or search set as the current selection. Identify it by name or guid (from list_selection_sets); organising folders are searched recursively. | ||
select_selection_set | Select the set called Mechanical Level 3 | Pro |
| Set the current selection to a specific set of items, addressed by their ids (from find_items, get_selection_tree, or get_current_selection). Replaces whatever is selected. | ||
set_current_selection | Select the items I just found | Pro |
Selection and Search Sets
| Tool | Example prompt | Edition |
|---|---|---|
| Save a property filter as a named search set. Unlike a fixed selection set, a search set re-runs its query so it stays up to date as the model changes. Give a name plus the category, property and value to match. | ||
create_search_set | Create a search set for all structural steel | Pro |
| Save a fixed list of items as a named selection set so it can be recalled later. Give the set a name and the ids of the items to include. | ||
create_selection_set | Save the current selection as a set called Level 3 Ducts | Pro |
| Delete a saved selection set or search set. Identify it by name or guid (from list_selection_sets). | ||
delete_selection_set | Delete the selection set called Old Test | Pro |
| List all saved selection sets and search sets in the document, including folders. Each entry reports its name, guid and kind (folder, search, or explicit); explicit sets also report how many items they hold. | ||
list_selection_sets | List the selection and search sets in this model | Free |
Clash Detection
| Tool | Example prompt | Edition |
|---|---|---|
| Add a comment to a clash result, optionally with an author and a comment status (new, active, approved or resolved). Identify the result by its name or its zero-based index within the test. | ||
add_clash_comment | Add a comment to clash 3 saying the duct needs dropping 200mm | Pro |
| Assign a clash result to a person or trade for action. Identify the result by its name or its zero-based index within the test. | ||
assign_clash_result | Assign clash 3 to the mechanical contractor | Pro |
| Create a clash test between two selections (a clash set). Each side is the name of an existing selection or search set, or a list of item ids. Choose the clash type and tolerance. | ||
create_clash_test | Create a hard clash test between Mechanical and Structure at 25mm | Pro |
| Export clash results as a BCF 2.1 file (.bcfzip) — one issue per clash, each with its camera viewpoint, a snapshot image, the assignee, status and comments. BCF is the open exchange format for coordination issues, so the result can be opened directly in Revit, Solibri, ACC and similar tools. Optionally filter by status so only the clashes the team agreed to action are exported. | ||
export_bcf | Export the active clashes in Mech vs Struct as a BCF file | Pro |
| Write a clash report for one test to CSV or HTML at the given path (the extension decides the format). Covers every result including those inside result groups, with status, distance, centre point, the two colliding items, assignee and description. The HTML report can embed a snapshot image per result. | ||
export_clash_report | Export the Mech vs Struct results to an HTML report with images | Pro |
| Navigate the camera to look at a clash result so the user can see it in the Navisworks view. Identify the result by its name or its zero-based index within the test. | ||
focus_on_clash | Take me to clash 3 in Mech vs Struct | Free |
| Save a snapshot image of one clash result to a PNG at the given path, rendered from the result's clash viewpoint. Choose the image size and style (scene, raytrace, or overlay with the clashing items highlighted). | ||
get_clash_result_image | Save a snapshot of clash 3 to my desktop | Pro |
| List the individual results of a clash test — each clashing pair with its status, distance, centre point and the two colliding items. Optionally filter by status and cap the count. | ||
get_clash_results | Show me the active clashes in Mech vs Struct | Free |
| Summarise one clash test: total number of results and a breakdown by status (new, active, reviewed, approved, resolved). | ||
get_clash_test_summary | Summarise the Mech vs Struct test by status | Free |
| Group clash results inside a test: create a named result group and move the listed results into it. Identify each result by its name or its zero-based index within the test's result list; entries that do not match are skipped and reported. | ||
group_clash_results | Group clashes 1 to 6 into a group called Level 3 Riser | Pro |
| List the ignore rules on a clash test with each rule's exact name and whether it is currently enabled. Ignore rules suppress uninteresting results (for example items in the same file or the same layer). Use the names reported here with set_clash_rules. | ||
list_clash_rules | What ignore rules are on the Mech vs Struct test? | Pro |
| List the clash tests in the document. For each test, report its name, clash type, status, number of results and tolerance. | ||
list_clash_tests | List the clash tests and how many results each has | Free |
| Run every clash test in the document, updating all results. Returns the number of tests run. | ||
run_all_clash_tests | Run every clash test in this model | Pro |
| Run one clash test by name, updating its results. Returns the test status and result count after the run. | ||
run_clash_test | Run the Mech vs Struct clash test | Pro |
| Set the review status of a single clash result (new, active, reviewed, approved or resolved). Identify the result by its name or its zero-based index within the test. | ||
set_clash_result_status | Mark clash 3 in Mech vs Struct as reviewed | Pro |
| Enable or disable ignore rules on a clash test, so a run stops reporting results the team does not want to see (for example clashes between items in the same file). Name rules exactly as list_clash_rules reports them. Re-run the test afterwards for the change to affect results. | ||
set_clash_rules | Turn on the rule that ignores items in the same file | Pro |
Sectioning
| Tool | Example prompt | Edition |
|---|---|---|
| Turn off all sectioning (box or planes) so the whole model is visible again. Reverses set_section_box and set_section_plane. Requires Navisworks 2025 or later. | ||
clear_sections | Turn sectioning off | Pro |
| Report whether sectioning is currently on, whether it is a box or planes, the box extents when in box mode, and which individual planes are enabled. Requires Navisworks 2025 or later. | ||
get_section_state | Is sectioning currently on? | Pro |
| Section the view to a box so everything outside it is cut away — the usual way to look inside a congested area such as a riser or plant room. Give an explicit min/max box, or set from_selection to box the current selection. Use clear_sections to remove. Requires Navisworks 2025 or later. | ||
set_section_box | Section a box around the current selection with 500mm padding | Pro |
| Section the view with a single plane aligned to a side (top, bottom, front, back, left, right or view), cutting away everything beyond it — use for a quick cut through a level or a wall line. Use clear_sections to remove. Requires Navisworks 2025 or later. | ||
set_section_plane | Cut a section plane through the top at level 12000 | Pro |
Appearance and Visibility
| Tool | Example prompt | Edition |
|---|---|---|
| Hide one or more items, addressed by their ids, so they no longer show in the scene. Use unhide_all to bring everything back. | ||
hide_items | Hide the selected items | Pro |
| Isolate a set of items: hide everything else in the scene and show only these items, addressed by their ids. Use unhide_all to restore full visibility. | ||
isolate_items | Isolate the current selection | Pro |
| Override the colour and/or transparency of one or more items, addressed by their ids. Give an RGB colour (each channel 0-255) and/or a transparency from 0 (opaque) to 1 (invisible). Use reset_appearances to clear. | ||
override_appearance | Colour the selected ducts red at 50% transparency | Pro |
| Clear every colour and transparency override in the scene, returning all items to their original appearance. | ||
reset_appearances | Reset all appearance overrides | Pro |
| Make every hidden item visible again, reversing any hide_items or isolate_items calls. | ||
unhide_all | Show everything again | Pro |
Viewpoints
| Tool | Example prompt | Edition |
|---|---|---|
| Move the current view to a saved viewpoint. Identify it by name or guid (from list_saved_viewpoints). | ||
apply_saved_viewpoint | Go to the viewpoint called Level 3 Riser Clash | Pro |
| Save the current view as a named viewpoint so it can be recalled later. Give the viewpoint a name. | ||
create_saved_viewpoint | Save this view as Level 3 Riser Clash | Pro |
| Delete a saved viewpoint. Identify it by name or guid (from list_saved_viewpoints). | ||
delete_saved_viewpoint | Delete the viewpoint called Old View | Pro |
| Read the current view's camera: its position, the direction it looks in, the up vector, and whether it is a perspective or orthographic projection. | ||
get_current_viewpoint | Where is the camera right now? | Free |
| List all saved viewpoints in the document, including the folders that organise them. Each entry reports its name, guid and kind (folder or viewpoint). | ||
list_saved_viewpoints | List the saved viewpoints | Free |
Viewpoint Animations
| Tool | Example prompt | Edition |
|---|---|---|
| Create a camera fly-through animation from an ordered list of saved viewpoints — the camera moves through them in the order given. Useful for a walkthrough of a coordination area or a review route. Note this is a viewpoint (camera) animation; object animation via the Animator is not scriptable, and rendering the result to video is done in Navisworks itself. | ||
create_viewpoint_animation | Build a fly-through through my four saved coordination views | Pro |
| Delete a camera fly-through animation by name, along with its frames. | ||
delete_viewpoint_animation | Delete the animation called Old Walkthrough | Pro |
| List the camera fly-through animations saved in this model, with how many viewpoint frames each holds and whether it loops. | ||
list_viewpoint_animations | List the animations saved in this model | Pro |
Navigation and Capture
| Tool | Example prompt | Edition |
|---|---|---|
| Save a snapshot of the current 3D view as a PNG image at the given path. Useful for sharing what the view looks like right now. | ||
capture_view_image | Save a screenshot of the current view | Pro |
| Move the camera to a position and aim it at a target point, keeping Z as the up direction. Coordinates are in the document's units. | ||
set_camera | Put the camera above the building looking down | Free |
| Zoom the current view to fit one or more items, addressed by their ids (from find_items, get_selection_tree, or get_current_selection). | ||
zoom_to_items | Zoom to the items I just found | Free |
| Zoom the current view to fit whatever is currently selected. | ||
zoom_to_selection | Zoom to what is selected | Free |
Measurement
| Tool | Example prompt | Edition |
|---|---|---|
| Measure the straight-line distance between two 3D points, with per-axis deltas. Coordinates are in the model's units. | ||
measure_points | How far is it between these two points? | Free |
4D Construction Sequencing
| Tool | Example prompt | Edition |
|---|---|---|
| Create a construction sequence (TimeLiner) task with planned dates and a task type (Construct, Demolish or Temporary), and optionally attach the model items it drives — either a named selection or search set, or explicit item ids. This is how a 4D sequence gets built. | ||
create_timeliner_task | Create a Construct task called Level 3 Ductwork starting 1 Sept for two weeks | Pro |
| Delete a construction sequence task by name. Removes the task and any sub-tasks beneath it. | ||
delete_timeliner_task | Delete the task called Old Sequence | Pro |
| Attach model items to an existing construction sequence task, replacing whatever it was linked to. Give either a named selection or search set, or explicit item ids. Use this to wire an imported programme up to the model. | ||
link_timeliner_task | Attach the Level 3 Ducts set to the Level 3 Ductwork task | Pro |
| List the simulation task types available in this model (typically Construct, Demolish and Temporary) with the appearance status each applies at the start and end of a task. Use these exact names for the task_type parameter of create_timeliner_task. | ||
list_timeliner_task_types | What task types are set up in this model? | Pro |
| List the construction sequence (TimeLiner) tasks with each task's name, status, planned/actual start and end dates, and task type. Walks the full task tree. | ||
list_timeliner_tasks | List the construction sequence tasks and their dates | Free |
Quantification
| Tool | Example prompt | Edition |
|---|---|---|
| Export the model's quantification (takeoff) data to a CSV for cost analysis in Excel — choose the takeoff objects (the measured quantities), the catalogue items, or the resources. Requires a quantification project that has already been set up and taken off in Navisworks; quantities cannot be created through the API. | ||
export_quantities | Export the takeoff quantities to a CSV on my desktop | Pro |
| Summarise the quantification (takeoff) project in the open model: currency and setup, counts of catalog items, item groups, takeoff objects, resources and steps, and a sample of catalog item names. Returns an error when the model has no quantification project. | ||
get_quantification_summary | Is there a quantification project in this model? | Free |
Review and Comments
| Tool | Example prompt | Edition |
|---|---|---|
| Add a review comment to a saved selection set or saved viewpoint, addressed by its display name (folders are searched recursively). Optionally set the author and a status. | ||
add_comment | Add a comment to the selected item | Pro |
| Read the review comments on a saved selection set or saved viewpoint, addressed by its display name (folders are searched recursively). Each comment reports its body, author, status and creation date. | ||
get_comments | Show the comments on this clash | Free |
Export and Publish
| Tool | Example prompt | Edition |
|---|---|---|
| Append (aggregate) another CAD, BIM or scan file into the current scene by path (e.g. NWC, NWD, RVT, IFC, DWG). Merges it with the models already loaded. | ||
append_model | Append the structural model into this federation | Pro |
| Write the model tree to a CSV (columns: Depth, Id, DisplayName, ClassName, HasGeometry), walking each model's hierarchy down to a maximum depth. | ||
export_model_tree | Export the model tree to a file | Pro |
| Save the scene as an NWC cache at the given path. NWC is the compact cache format used to aggregate models into a coordination scene. | ||
export_nwc | Export the current model to NWC | Pro |
| Save the whole scene to an NWD file at the given path. An NWD bundles all geometry and review data into a single self-contained file you can hand off. | ||
export_nwd | Export this federation as an NWD | Pro |
| Save the scene to an NWF file at the given path. An NWF references the appended source files rather than embedding their geometry, so it stays small and updates when the sources change. | ||
export_nwf | Save this as an NWF | Pro |
| Write the property categories and values of items to a CSV (columns: Item, Category, Property, Value). Uses the given item ids, or the current selection when no ids are supplied. | ||
export_properties | Export the properties of the current selection | Pro |
| Publish the scene to a read-only NWD file with embedded title, author and comments. Use for distributing a locked, self-contained review deliverable. | ||
publish_nwd | Publish an NWD with the title set to Level 3 Coordination | Pro |
Troubleshooting
The AI says it is not connected to Navisworks
- Navisworks must be running with a drawing open. The connector talks to a live session; it cannot open Navisworks for you.
- In Navisworks, open MCP Status from the Tool add-ins tab to check the plug-in state, tool count and pipe status.
- Ask your AI to run the
pingtool. It reports whether the link is up and, if not, why.
The plug-in did not load
- The plug-in auto-loads when Navisworks starts. If the AUTOM8LABS tab is missing, restart Navisworks and check the startup log at
%LOCALAPPDATA%\AUTOM8LABS\MCPConnectorNavisworks\Log\. The connector's commands also remain available under the Tool add-ins tab as a fallback. - Confirm the bundle exists at
C:\ProgramData\AUTOM8LABS\MCPConnectorNavisworks. If it is missing, re-run the installer. - Check that your Navisworks release is between 2022 and 2027.
Pro tools are missing after activating
- The MCP server reads the edition once, when the AI client starts it. Restart the AI client after activating and the full tool list appears.
- Confirm the licence state in Navisworks via Manage Licence on the AUTOM8LABS ribbon.
The AI client does not list the server
- Restart the client completely after installing; a window reload is not always enough.
- For clients other than Claude Desktop, ChatGPT Desktop, and Cursor, add the manual configuration yourself.
FAQ
Autodesk Navisworks 2022 through 2027 on Windows. One install covers all supported releases.
No. Navisworks LT does not support add-ins, and the connector runs as a plug-in inside Navisworks. Full Navisworks 2022 to 2027 is required.
No. The free edition works without any key. It gives you the read-only inspection tools: query the drawing, list its tables, read properties, measure, and zoom.
Buy a Pro subscription on the product page (£10 per seat per month or £100 per seat per year). Your key arrives by email. Activate it in Navisworks via Manage Licence on the AUTOM8LABS ribbon or Manage Licence under the Tool add-ins tab, then restart your AI client.
Yes. A seat is per machine. Click Deactivate in the licence dialog to free the seat, then activate the same key on the new PC.
The free edition always works offline. A validated Pro key is honoured offline for 14 days, so short offline stretches never block your work. Note that your AI client may still need its own connection to reach its model provider.
The connector itself runs entirely on your machine: the MCP server and the Navisworks plug-in talk over a local connection. Tool results are returned to your AI client, which handles them under its own provider's terms, so only the data a tool returns ever leaves the connector.
No. There is no raw command execution tool in MCP Connector for Navisworks - the AI can only call the structured tools listed in the Tool Reference. Every tool that changes the model runs inside a Navisworks transaction, and the free edition cannot modify anything at all.
Yes. Each client launches its own copy of the MCP server, and one Pro seat covers every AI client on the machine.