brainless / dwata Goto Github PK

As a user I would like updated list of the AI models that are supported, including latest versions and costs. Right now list of models is in Rust code and this should be moved into SQLite DB which can sync from an online list (on https://dwata.com).

As the list of supported models and AI providers increase, and many providers support using OpenAI compatible API, we can also add support for new models without changing code.

As a user I would like Dwata to provide AI models with a list of functions which provide context to my chat. Functions would include:

• Name (with optional path) of CSV file(s)
• Schema of files

As a user I would like Dwata to provide AI models with a list of functions which provide context to my chat. Functions would include:

• Name and engine of database
• List of tables
• Schema of tables
• Result from execution of SQL query

As a user I would like to have a sample database to try out Dwata. This could be real data that is interesting to a niche audience, like:

• List of incubators, accelerators, angel investors - this would be useful for founders
• Sample eCommerce company - this would be more generic but will be less directly useful by itself

As a user I would like to be able to use Anthropic's AI models like Claude 3 with Dwata.

As a user I would like to have pricing information when selecting an AI model. This will need to be stored per model and shown in the UI where I select AI model for chat.

As a user of Dwata, I would like to be able to connect to SQLite databases, read schema and data from them. i would like Dwata to provide schema information in function callbacks in AI chats.

As a user of Dwata, I would like to be able to connect to a MySQL databases, read schema and data from them. i would like Dwata to provide schema information in function callbacks in AI chats.

As a user I would like to import Markdown files, for example from my company documentation. A folder containing Markdown files should become a data source.

As a user I would like to access Google Gemini AI models through Dwata.
https://ai.google.dev/docs

As a user I would like to set a folder as a source of data. These are the rough criteria:

• Find and import files of type text and Markdown (#101) and CSV (#89)
• Folder should be scanned recursively
• Ability to specify a depth
• Ability to specify exclude patterns (check how IDEs work)

As a user of Dwata I would like to able to read both data and structure of CSV files.