Enabling Codex to search through multiple repo’s before giving an answer would allow users to get much better solutions and answers about complex coding problems that require you to work with multiple tools.
For instance:
If you work with a database to store data you ingest from a data streaming platform – knowing which data types are supported on both ends, the specs of interfaces on each side and the syntax of the scripting languages both use, would allow you to much faster prototype a working data pipeline.
Sure, there are workarounds by just merging repo’s on your end or using submodules and subtrees in Github, but if Codex would automate that process it would be much more convenient for end users and provide a substantial efficiency gain!
Like, comment, bookmark or add this thread to your watchlist to make this a priority for OpenAI!