Conversation
NA-Wen
left a comment
There was a problem hiding this comment.
Thanks for the contribution — this workflow looks very interesting.
One question: currently the visualizer agent outputs a Mermaid mind map code block. For many users, it might be more convenient if the workflow could directly provide a rendered image of the mind map, instead of requiring users to render the Mermaid code themselves.
Would it be possible to modify the workflow so that the visualization step outputs a rendered diagram (e.g., PNG/SVG) rather than raw Mermaid syntax?
This would make the workflow easier to use and more intuitive for users who just want to quickly view the visualization.
|
My 2 cents: what about allowing the mermaid to be rendered on the frontend? |
Good call. We could consider rendering Mermaid on the frontend. I’ll check what the implications would be 😃. |
Created a new workflow which can be used to summarize and visualize articles. It could be applied to argumentative articles, academic researches and other articles that are logically structured. It includes three agent nodes. The summarizer drags basic contents of the article it reads; the writer turns fragmented points into a fix-structured review; the visualizer gives out a mermaid code that shows a mindmap of the article based on the writer's output.