How to work with monorepos and how to use PAELLADOC correctly?

[VladimirLevadnij]

Hi, @jlcases ! Thank you for your work!

I'm trying to understand how PAELLADOC works with monorepos and how to work with PAELLADOC correctly 😊

I want to understand using the example of the Flowise monorepo, on the basis of which I am making part of my project https://github.com/FlowiseAI/Flowise

Initially, I thought that the commands for activating PAELLADOC should be executed directly in the terminal, but from the command I realized that it is not for the terminal, and as I understand it, these commands are for the form of communication with the agent in Cursor? Did I understand this correctly?

I go this way:

1. First, I cloned the repository https://github.com/jlcases/paelladoc then created and activated a virtual environment, and then installed dependencies, but I did not understand why I did this?

Questions 1: Can you explain why this step (installation of dependencies) is needed? Is this for developing the PAELLADOC framework itself, not for using it? Can I just skip this for using it?

2. Then I select the Agent mode in Cursor, select a neural network, for example claude-3.7-sonnet and enter a command, for example, GENERATE_CONTEXT repo_path=/home/name/Projects/Flowise and the job starts, which creates the extracted documentation in paelladoc/docs/Flowise and duplicates it in paelladoc/.cursor/project_context.txt and in paelladoc/.cursor/project_context.json.

Questions 2: Is everything correct, is it supposed to be like this? Which AI model is better to choose, claude-3.7-sonnet or another one? Will the result be better if you enable the thinking mode in claude-3.7-sonnet or another model?

3. Then in Cursor, being in the paelladoc project NOT IN THE TERMINAL, but in the request chat for the neural network, I entered the command GENERATE_DOC repo_path=/home/name/Projects/Flowise and the work began.

Questions 3: Since my default language is Russian, I was asked to create documentation in "Russian (ru)", do I understand correctly that I can create documentation in different languages, for example, in Russian and English, and to provide context for the agent's work in Cursor, provide it with documentation in English, but at the same time, there will be documentation in Russian?

I already have an existing Flowise project, which documentation options should I choose? I decided to select item "19. Everything". Did I do the right thing, or would it have been better to do it separately?

Even though I wrote repo_path=/home/name/Projects/Flowise the documentation was generated in paelladoc/docs/generated/... is it supposed to be like that or did something not work?

Now I need to copy this documentation to the Flowise/.cursor/rules folder myself?

These are the first questions I had when I started working, but the MAIN QUESTION is, in case of using a monorepo, which I work with as a single whole, do I understand correctly that I need to create centralized documentation for the entire monorepo, and not separate for each part of the monorepo?

Thank you, I hope for your answers 😊

[VladimirLevadnij]

Based on the information in https://github.com/jlcases/paelladoc/blob/main/CHANGELOG.md it turns out that "Generated files are now always saved in /docs/generated/ while following template structure".

So, after all, the GENERATE_DOC repo_path= structure is obsolete and does not work? Or did I need to select output=~/custom/docs to select a different output folder?

And, as I asked above, I do not yet fully understand the recommended further steps, do I need to copy this documentation to Flowise/.cursor/rules and somehow enable it through the Cursor settings or give a link to this documentation when creating a new request when working with the Agent?

[VladimirLevadnij]

Do I need to create such a structure myself?

Because at the moment PAELLADOC has not created such a structure, it has created this structure.

[jlcases]

Hi @VladimirLevadnij!

Thank you so much for your detailed questions and for trying out PAELLADOC! You've raised some excellent points about monorepo support and the overall workflow. Let me address your concerns:

1. About Dependencies & Setup

   • You're right - the current setup process isn't clear enough. The dependencies are primarily for development. For just using PAELLADOC with Cursor, you shouldn't need to clone the repo or install dependencies. We'll improve the documentation to make this clearer.

2. About AI Models & Output

   • Claude-3.7 is indeed a good choice for PAELLADOC
   • The multiple output locations you're seeing (paelladoc/docs/Flowise, .cursor/project_context.*) are a bug we're fixing
   • The intended behavior is to have a single, clear output location

3. About Multilingual Support

   • Yes! You can generate documentation in multiple languages form the last pull request
   • The context/analysis is language-agnostic
   • Each GENERATE_DOC run can target a different language

4. About Monorepo Support (your main question)

   • Currently, PAELLADOC works best with a centralized documentation approach for monorepos
   • However, this is an area we're actively working on improving
   • We plan to add better support for per-package documentation while maintaining cross-package relationships

5. About Directory Structure

   • The inconsistency you're seeing between expected and actual directory structure is a known issue
   • We're working on making the output structure more consistent and configurable

I'll be improving the documentation and fixing these issues in the coming days. The project is still in early stages, and feedback like yours is incredibly valuable.

Would you be interested in helping us improve the monorepo support? Your experience with Flowise could provide great insights.

Thanks for your patience and support! 🚀

[VladimirLevadnij]

@jlcases Hi! Thank you for your detailed answer! At the moment I am experimenting with Memory Bank System v0.6-beta https://github.com/vanzan01/cursor-memory-bank

I see that PAELLADOC has even more potential, I am waiting for new versions of PAELLADOC to check and think how I can use PAELLADOC and maybe how I can help in development 😁