I recently discorve codespell and it was a great one, particularly in so far as I am not fluent in English.

However, after experimenting on various projects, I strongly suggest adding a feature per-file-ignores to ignores specific words into specific files. This would give finer control over the errors returned by codespell, as we sometimes use some variable names which are flag by codespell, but we don't want to exclude them globally they could really be a legitimate typo error in others context.

The toml configuration would look like this:

[codespell]
[codespell.per-file-ignores]
"file1" = "word1,word1"
"file2" = [
  "word3",
  "word4",
]

The inline/command format will be a bit more like that:

codespell  --per-file-ignores "file1:[word1,word2],file2:[word3,word4]"