Control the DFCine Canvas with WorkBuddy

After completing this guide, you can use natural language in WorkBuddy to read and operate the canvas currently open in DFCine. You can also configure a WeChat bot and control the canvas from a WeChat conversation.

Before You Begin

Make sure that:

• DFCine is installed and activated.
• A project and canvas are open in DFCine.
• You are using Windows and can open PowerShell.
• You do not share your local DFCine control token or complete configuration script with anyone.

Step 1: Download and Install WorkBuddy

Visit the WorkBuddy invitation and download page:

1. Click Download WorkBuddy.
2. Select the Windows x64 version on the download page.
3. Open the installer and follow the instructions.
4. You can skip connector setup for now and continue with the DFCine configuration below.

Step 2: Generate the WorkBuddy Configuration in DFCine

1. Open DFCine and enter a project canvas.
2. Open DFCine Settings.
3. Select MCP Agent.
4. Check whether Local Control Token contains a value.

If the token is empty, click Regenerate Token.

If Codex is already configured and can use DFCine MCP, do not regenerate the token. Regenerating it immediately invalidates the old token, so the existing Codex configuration must be updated by running the Codex script again.

Find the WorkBuddy MCP Configuration section:

1. Click Copy Script next to the complete configuration script.
2. Press the Win key, search for PowerShell, and open it.
3. Paste the complete script into PowerShell.
4. Press Enter to run it.

When the command succeeds, PowerShell confirms that the DFCine MCP configuration was written to the WorkBuddy configuration file and asks you to restart WorkBuddy.

The script preserves existing WorkBuddy MCP services and only adds or updates the server named dfcine. You do not need to edit configuration files manually or set system environment variables.

Return to Settings > MCP Agent in DFCine:

1. Confirm that Enable is selected.
2. Click Save MCP Settings.
3. Keep DFCine, the current project, and the canvas open.

The default local service address is http://127.0.0.1:38765/mcp. You normally do not need to change the local port.

Step 3: Enable the DFCine Connector in WorkBuddy

If WorkBuddy is already open, exit it completely and reopen it so that the new configuration takes effect.

1. Open WorkBuddy.
2. Select Connectors in the conversation window.
3. Click Manage Connectors.
4. Click Custom Connector in the upper-right corner.
5. Find dfcine under My MCP.
6. Click Trust when prompted the first time.
7. Confirm that dfcine is enabled.

If dfcine appears under My MCP, WorkBuddy has loaded the DFCine MCP configuration.

Step 4: Test the WorkBuddy Connection

Create a new task in WorkBuddy:

1. Set the task mode to Plan.
2. Select Auto or another model.
3. If needed, connect a third-party model by following the WorkBuddy documentation.
4. Send a test instruction.

Start with:

Can you access the DFCine MCP features? List the DFCine tools that are currently available.

When the connection is working, WorkBuddy returns tools whose names begin with dfcine_. You can also enter:

Read the current DFCine canvas information and tell me which node is selected.

If WorkBuddy returns the current canvas or node information, the configuration is complete.

Step 5: Control the Canvas with WorkBuddy

You can now describe the operation you want WorkBuddy to perform. For example:

Read the selected DFCine node and its workflow instructions. Optimize the prompt for the current workflow, show me the planned changes first, and then write them to the node.

Or enter:

Read the node list on the current canvas and explain which MCP operations are available for each node.

WorkBuddy can only use the MCP features currently exposed and authorized by DFCine. Before an operation writes parameters or runs a node, review the plan and confirmation information shown in the task.

Step 6: Control DFCine from WeChat

WorkBuddy can also receive tasks through a WeChat bot. See the WorkBuddy WeChat ClawBot setup guide for configuration instructions.

After binding WeChat and configuring the bot, you can:

• Send canvas control instructions to WeChat ClawBot from a WeChat conversation.
• Select the model for WeChat control tasks from the Assistant button in WorkBuddy.
• Ask WorkBuddy from WeChat to call the enabled DFCine MCP tools.

The computer running WorkBuddy and DFCine must remain powered on. DFCine must have a project and canvas open, and MCP Agent must remain enabled.

Troubleshooting

dfcine Does Not Appear in WorkBuddy

• Confirm that you ran the complete script from WorkBuddy MCP Configuration in PowerShell.
• Exit WorkBuddy completely and reopen it.
• Check Connectors > Manage Connectors > Custom Connector > My MCP again.
• Return to DFCine, copy the WorkBuddy script again, and rerun it.

dfcine Appears but Its Tools Are Disabled

• Click Trust when using it for the first time.
• Confirm that the dfcine connector switch is enabled.
• Create a new task for testing instead of continuing a task opened before configuration.

The Tools Appear but Cannot Read the Canvas

• Confirm that DFCine is running.
• Confirm that Enable is selected under MCP Agent and that Save MCP Settings was clicked.
• Confirm that a project and canvas are open in DFCine.
• Confirm that WorkBuddy uses the same port and token currently configured in DFCine.

Codex Stopped Working After WorkBuddy Was Configured

The WorkBuddy configuration does not overwrite the Codex configuration. The most common cause is regenerating the token during setup.

If you regenerated the token, copy and run both of these scripts again:

• The complete script under Codex Global Integration.
• The complete script under WorkBuddy MCP Configuration.

Restart both Codex and WorkBuddy afterward.

Connection Failed After Changing the Port or Token

After changing the port or token, run the latest WorkBuddy configuration script and restart WorkBuddy. If Codex is also configured, run the latest Codex script and restart Codex as well.