http is already registered as a protocol and you can indeed do that already. It's printed on every table page. eg on https://csvbase.com/calpaterson/opcodes-6

http is already registered as a protocol and you can indeed do that already. It's printed on every table page. eg on https://csvbase.com/calpaterson/opcodes-6502 https://csvbase.com/calpaterson/opcodes-6502: import pandas as pd; opcodes_6502 = pd.read_csv('https://csvbase.com/calpaterson/opcodes-6502', index_col=0) But writing back is harder. It's not easy to make pandas do an HTTP PUT and then insert the HTTP basic auth and so on. Plus (not discussed in the article) there is a cache to avoid redownloading the data when it hasn't changed, which is my personal bete noire in "data science" such as it is.