Dynamo MCP

MCP stands for Model Context Protocol. It is an open standard that lets AI models connect to software through a common protocol. An MCP server exposes a program's tools and data so an AI client can read live context, take actions, and return structured results.

DynamoMCP is Dynamo's built-in MCP server. It connects Autodesk Assistant to your active Dynamo session — your graph, nodes, wires, and workspace state — so AI help is grounded in what you are actually working on.

For Dynamo users, that means Assistant can do more than answer general questions. It can inspect your open graph, help troubleshoot node issues, and update the canvas based on real session context instead of manual copy and paste.

DynamoMCP is available in Dynamo 4.2.0 and later.

How to Use DynamoMCP

You use DynamoMCP through Autodesk Assistant — there is no separate setup or standalone interface. Assistant is the entry point for asking questions, getting help with your graph, and letting the AI work with live Dynamo context.

Where you open Assistant depends on where you are working:

• Dynamo Sandbox — Dynamo MCP tools are available in Autodesk Assistant inside Dynamo.

• Revit — Dynamo MCP tools are available in Autodesk Assistant inside Revit.

In both cases, you stay in your current session and use the same Assistant experience to reach DynamoMCP.

What DynamoMCP can do

DynamoMCP gives Assistant a set of tools. Each tool does one job in Dynamo, such as reading the canvas, placing a node, or running the graph. You never call these tools yourself. You describe what you want in plain language, and Assistant picks the tools it needs.

Knowing what the tools are is still useful. It tells you what Assistant is able to do, and it helps you phrase requests it can act on.

Tool names that start with get_ only read information from your session. Every other tool can change your graph.

Read the current graph

These tools let Assistant see what you are working on before it answers or makes a change.

Tool	What it does
get_workspace_info	Lists what is on the canvas: nodes, wires, groups, notes, and workspace details.
get_selection	Reports the nodes, wires, and groups you have selected.
get_node_connections	Shows what a node is wired to, port by port.
get_node_property	Reports node settings such as preview, freeze, pinned preview, and labels.
get_node_lacing	Reports the lacing setting of a node.
get_installation_info	Reports your Dynamo version, the host you are running in, and a link to the release notes.

Example prompt: "What is on my canvas right now?"

Find the right node

Dynamo has thousands of nodes, including nodes from installed packages. These tools help Assistant choose the correct one instead of guessing.

Tool	What it does
get_nodes_info	Searches the node library, including package nodes, and returns the exact node to place.
get_node_help	Returns the written help for a node, and the path to its example graph when one exists.
get_type_schema	Returns the expected data format for a Dynamo type, such as Point, Number, or Color.

Example prompt: "Which node do I use to get the area of a surface?"

Build and edit the graph

These tools do the work on the canvas.

Tool	What it does
create_nodes	Places one or more nodes on the canvas, with optional starting values.
set_node_value	Sets the value of input nodes such as Number, String, Code Block, and dropdowns.
connect_nodes	Wires an output port to an input port.
disconnect_nodes	Removes wires between two nodes.
delete_node	Removes a node from the canvas.
rename_node	Renames input and Watch nodes so their purpose is clear. Function nodes keep their original names.
select_nodes	Selects nodes on the canvas, so you can see which ones Assistant means.
set_node_property	Turns node settings on or off: preview, freeze, pinned preview, labels, and input or output.
set_node_lacing	Sets lacing to Auto, Shortest, Longest, or Cross Product.
set_node_port_list_at_level	Sets the list level on an input port.

Example prompt: "Add a number slider and wire it into the radius input of my circle."

Organize and document the graph

A working graph is not always a readable one. These tools handle layout, grouping, colors, and notes.

Tool	What it does
create_group	Puts a group around a set of nodes.
create_groups	Creates several groups in one step.
ungroup_by_guid	Removes a group. The nodes inside it stay on the canvas.
ungroup_by_title	Removes groups by their title. The nodes inside them stay on the canvas.
get_available_group_styles	Lists the group styles you can use, including custom styles from your preferences.
assign_group_styles	Applies group styles, such as Inputs, Actions, Outputs, and Review.
get_notes	Lists the notes in the graph.
create_notes	Adds notes that explain parts of the graph.
update_notes	Edits the text of a note, or moves it.
delete_notes	Removes notes.
set_note_state	Pins a note to a node so the two move together, or unpins it.
auto_layout_workspace	Rearranges the nodes using the built-in Dynamo auto layout.
set_workspace_description	Writes a description of the graph into the file's properties.

Example prompt: "Group and color-code this graph by inputs, actions, and outputs, then add short notes."

Run the graph and check results

Tool	What it does
run_workspace	Runs the current graph. In automatic mode the graph already re-runs on every change, so this step is skipped.
get_workspace_warnings	Lists the nodes with warnings or errors, the message text, and suggested next steps. The graph must run first.
get_node_output_values_shapes	Summarizes the output values of one node, or of every node in the graph.
get_node_output_value	Reads the full output of a single node, one page at a time, for results that are too large to summarize.

Example prompt: "Run the graph and tell me why the List.GetItemAtIndex node is showing a warning."

Work with graph files

Assistant can also work with graphs on disk, not only the one you have open.

Tool	What it does
new_workspace	Opens a new, empty graph. If the current graph has unsaved changes, you are asked to save or discard them first.
save_workspace	Saves the current graph. Without a location, it saves to your Dynamo folder in Documents.
edit_graph_from_path	Opens a saved graph for editing, without running it.
insert_graph_from_path	Merges a saved graph into your current canvas. This is how node example graphs are brought in.
run_graph_from_path	Runs a saved graph, with optional input values that you supply.
run_graph_from_json	Runs a graph supplied as Dynamo JSON, with no file on disk.
get_graph_info_from_path	Reads the inputs, outputs, and metadata of a saved graph without opening it.
get_graph_info_from_json	Reads the same details from a graph supplied as Dynamo JSON.
get_graphs_info_from_path	Scans a folder of graphs and reports what each one contains.
get_available_graphs_folders	Lists the folders Dynamo knows about that may contain graphs.

Example prompt: "Look through my Dynamo graphs folder and tell me which graphs place walls."

Assistant can change your graph. Save your work before you ask for a large edit, and review the result before you save over an existing file.