{"version":"2.1.296","anchor":"non-utf-8-files-are-refused-by-notebookedit-instead-of-being","canonical_anchor":"non-utf-8-files-are-refused-by-notebookedit-instead-of-being","heading":"NotebookEdit refuses notebooks that are not UTF-8 instead of quietly rewriting them","tier":"notice","area":"Notebooks","scope":"individual","heads_up":true,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.296\/e\/non-utf-8-files-are-refused-by-notebookedit-instead-of-being","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.296","markdown":"### NotebookEdit refuses notebooks that are not UTF-8 instead of quietly rewriting them\n\nNotebookEdit now refuses to edit a notebook that is not valid UTF-8 and leaves the file untouched\n\n**What**\n\nUTF-8 is the standard way of storing text as bytes. Some files use an older encoding (such as Windows-1252, Shift-JIS or GBK) or are not text at all. When NotebookEdit, the tool Claude uses to edit Jupyter notebooks, finds a notebook like this, it now stops and writes nothing. Before, it decoded the file as best it could, which can garble characters, and saved the result as UTF-8.\n\nThe error tells Claude that the file is not valid UTF-8 and suggests using a shell command or asking you about converting the file.\n\nThere is also a similar message for Write. It covers new content that contains the replacement character (U+FFFD, the symbol that stands in for text that could not be read) when the file already on disk is not valid UTF-8.\n\n**Why**\n\nFiles saved in an older encoding are no longer silently damaged when Claude edits them. You get a clear refusal instead and can decide how to handle the file.\n\n- Area: Notebooks\n- Tier: You'll notice\n- Useful: 3\/5\n- Signal: 1\/5\n- Scope: individual\n- Heads-up: yes"}