What would you like to see?

Currently, we support the uploading of documents and links via the collector endpoints in the API. There is a use case though for an API user to directly send raw text and preferred metadata directly to the collector for it to then format and save.

note the metadata cannot allow arbitrary key values or else database schemas for many vector dbs will prevent the upsert as sometimes the metadata is spread into columns and the columns must be homogenous across that collection.