🫱🏻‍🫲🏼📚✨ First LLM AI files released!

[Dean D. Babic]

Hi all,

I checked all Docs and just released first versions of this two files:

https://jampy-docs-v7.readthedocs.io/en/latest/llms-full.txt

https://jampy-docs-v7.readthedocs.io/en/latest/llms.txt

Here is the ChatGPT chat used with this file:

https://chatgpt.com/share/684ebdd5-3f74-8012-b42a-565ebb19315c

You're very welcome! I'm glad it was helpful — and that's great to hear you're sharing it publicly! 🚀

If anyone else reading this has questions about Jam.py, low-code frameworks, or Python web development, feel free to reach out. Jam.py is a powerful tool once you understand its flow and schema-based design.

Good luck with your project — and thank you for contributing to the open-source and knowledge-sharing community! 🫱🏻‍🫲🏼📚✨

Enjoy

[Dean D. Babic]

Ha!

Looks to me that NONE of the AI's can fetch the file from readthedocs.

For example, ChatGPT:
You're right to raise that point — I don’t have built-in knowledge of the entire llms-full.txt from Jam.py version 7, and due to a server error (HTTP 500), I couldn’t retrieve its latest version directly from ReadTheDocs just now.

And Gemini:
I am sorry, but I am still unable to access the content of the webpage you provided at https://jampy-docs-v7.readthedocs.io/en/latest/llms-full.txt. This could be due to various reasons such as paywalls, login requirements, or the content being sensitive.

And here is a catch! If I try to upload it, the file for Gemini will be:

The uploaded file llms-full.txt contains 475 lines of content. 

And GPT:
The llms-full.txt file you provided contains approximately 8,891 lines.

Hence, both will be waaay less than the original. So for GPT we can upload a zip file.
Gemini wants Google Drive, as txt file. But it reads way less than GPT.

Claude.ai did not preform at all.

Both Ai's will try to reference jam-py.com/Docs as official site AND jam-py.readthedocs.
And there, the Docs are for V5, and not corrected.

D.
 

[Dean D. Babic]

ChatGPT is waaay ahead! Just feed it with the llms-full.txt as a zip file.

It now recognizes everything and it is able to scaffold the Jam.py app from scratch. It is still using v5 "engine" schema (because I'm lazy ;).

I just asked: "Hi, would you know now how to "scaffold" the Jam.py app from database schema?"

It still does not understand that EACH NEW Jam project comes with predefined Journals, Catalogs and Reports.

Hence, if someone is keen to try to scaffold a new app, pls feed it with defaults (tell it to use it):

SYS_ITEMS config:
ITEM_START_ID = 6
PARENT_ID = 2
TASK_ID = 1
TYPE_ID = 10
VISIBLE = 1
DELETED = 0
TABLE_ID = 0

SYS_FIELDS config:

FIELD_START_ID = 15
OWNER_ID = 3
F_ALIGNMENT = 1
F_TEXTAREA = 0
F_DO_NOT_SANITIZE = 0
F_CALC_LOOKUP_FIELD = 0
F_REQUIRED = 0

As in here: https://github.com/jam-py-v5/gnucash-2-web/blob/main/utils/ai8.py

This is soooo fun. There is way more, of course, like reserved words and non English language.
Like, we can't blindly use any language for Name* in Builder.
It must be JavaScript valid variable with no spaces, etc.

ChatGPT CAN reference previous chat. Just ask for it if it starts rubbishing.

D.

[Dean D. Babic]

Hi all, 

looks like this file:

https://jampy-docs-v7.readthedocs.io/en/latest/llms-full.txt

Does not contain full path to the images used in Docs.

The AI CAN recognize what is in the image, it can do the OCR. Try it!

However, with no correct path, it will never learn what the image represents.

Trying to fix this with the developer here:

https://github.com/jdillard/sphinx-llms-txt/issues

No much luck yet... 

[Ari Sobel]

Dear Drazen,

Congrats!

I am so out, that I don't even understand what is intended here. 

Can you explain, please?

thanks!

Ari

--
You received this message because you are subscribed to the Google Groups "Jam.py Users Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jam-py+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/jam-py/59130862-c411-41e3-b4d9-2923037517e8n%40googlegroups.com.

[Dean D. Babic]

This is for AI engines Ari.

It is possible to copy/paste this file into the ChatGPT so it can LEARN!

Or, some engines crawl this file so they would pick it up from  jam-py.com

If, and only if the file exist there.

[Dean D. Babic]

Here you go! 

All images fixed! For V7 only.

[Dean D. Babic]

So Ari, 

just upload this file to ChatGPT and ask any question on your language. 
The file is also here:
https://jampyapplicationbuilder.pages.dev/llms-full.txt

Just ask to refer this file, otherwise it will hallucinate.

I asked "explain routing in jam.py":

Routing u Jam.py V7

“The Jam.py v7 introduced routing. As Jam.py v5 is a SPA (Single Page Application), there was no need for routing.”

...
It will still try to help with wrong info, like SYS_PAGES i SYS_ROUTING - that does not exist. 

  

D.

[nur...@gmail.com]

Well this is neat. Ill have to see if i can reformat it to use in RAG ( remember, not everyone uses those commercial LLMs, for a variety of reasons )

[Dean D. Babic]

The V7 Docs are here:

https://github.com/jam-py-v5/jampy-docs-v7

Install as per requirements.txt 

and just run:

sphinx-build . html 

That WILL create llm files as well. 

If there are some typos or errors, pls do the pull request.

The Docs are for you guys, only you people can make it better.

It is now possible to create a complete app with just a script:

https://asciinema.org/a/lG3VLqhQtsmUDK1EhYxLIzTI3

We can now hit MySQL and get a complete app in a second:

python3 ai_mysql.py
usage: ai_mysql.py [-h] --database DATABASE --user USER --password PASSWORD [--host HOST] [--port PORT] [--encoding ENCODING]

The result is similar to above.

The next stage would be to create lookups automatically.
However the main issue is a lack of Primary Keys on existing apps for this to happen.

D.