I find myself asking ChatGPT to generate a lot of mermaid structures, and I would find it very helpful if this client at least had an opt-in feature wherein multiline code blocks of the format that follows would, if valid, render as the structure they are written to represent.
As you can see, github renders this graph, making it easy to visualize immediately.
Considerations
It would be important to include a button to toggle between the rendered mermaid structure and displaying the raw code itself -- here is an example from the Markdown live editor obsidian.md
On error, as of course ChatGPT will make errors in generating these diagrams--fairly frequently in my experience actually--the parse error could be shown to the user.
this is again inspired by obsidian's implementation, which I believe is very good.
Motivation
This feature would help me more quickly interpret the verbose output of the ChatGPT simply by instructing it to use these visualization tools in its output, providing me with a more feature-rich, conducive-to-learning experience while using the LLM chat agent.
Alternatives
Alternatively, this could certainly be:
OPT-IN - as I mentioned earlier, it is not necessary that this be a feature that everyone uses, and can be a default: off setting if that is preferred.
LEFT OUT - This is obviously just an enhancement, and it could be omitted without causing real damage to myself or others. I simply would very much appreciate the feature.
Additional context
This feature is fairly niche, but it could reasonably be implemented in a way which gives rise to ease-of-additional functionality in the way the front matter is rendered to the user, which I believe would be an excellent selling point of the desktop application over simply using the browser. Besides syncing prompts, of course.
Feature description
Overview
I find myself asking ChatGPT to generate a lot of
mermaid
structures, and I would find it very helpful if this client at least had an opt-in feature wherein multiline code blocks of the format that follows would, if valid, render as the structure they are written to represent.Here is an example, using a
graph
.Considerations
It would be important to include a button to toggle between the rendered
mermaid
structure and displaying the raw code itself -- here is an example from the Markdown live editor obsidian.mdOn error, as of course ChatGPT will make errors in generating these diagrams--fairly frequently in my experience actually--the parse error could be shown to the user.
Motivation
This feature would help me more quickly interpret the verbose output of the ChatGPT simply by instructing it to use these visualization tools in its output, providing me with a more feature-rich, conducive-to-learning experience while using the LLM chat agent.
Alternatives
Alternatively, this could certainly be:
default: off
setting if that is preferred.Additional context
This feature is fairly niche, but it could reasonably be implemented in a way which gives rise to ease-of-additional functionality in the way the front matter is rendered to the user, which I believe would be an excellent selling point of the desktop application over simply using the browser. Besides syncing prompts, of course.