studysolo-workflow-canvas

To safely edit workflow canvas nodes, retrieve the latest nodes_json and edges_json via MCP tools, translate changes into apply_workflow_canvas_patch operations, and use dry-run validation to prevent overwriting persisted graph state.

You can apply multi-node modifications to a DAG graph by reading the current canvas and node manifests, generating apply_workflow_canvas_patch operations for the new nodes and edges, and validating the changes with a dry run before persisting.

Yes, after a successful dry-run validation of your canvas patch, you can execute the updated workflow immediately using the run_workflow_and_wait MCP tool, which also provides SSE execution trace data for monitoring.

If the graph is updated concurrently, the system handles conflicts by re-fetching the canvas to confirm the latest persisted state, ensuring your node modifications do not unintentionally overwrite recent changes made by others.

No, this approach specifically maintains existing workflow instances by editing real persisted node instances and edges, rather than supporting the development or mocking of entirely new node types.

To confirm your canvas edits are persisted, re-read the workflow canvas via MCP tools after applying the patch to verify the updated nodes_json and edges_json reflect your changes and check the updated_at timestamp.