{"version":"2.1.318","anchor":"read-tool-wraps-saved-copies-in-an-envelope","canonical_anchor":"read-tool-wraps-saved-copies-in-an-envelope","heading":"The Read tool can wrap saved file copies in marking text","tier":"internal","area":"Read Tool","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.318\/e\/read-tool-wraps-saved-copies-in-an-envelope","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.318","markdown":"### The Read tool can wrap saved file copies in marking text\n\nWhen a saved copy is set to be enveloped, the Read tool now wraps the file's text in framing and neutralizes any lookalike tags inside it\n\n**Unclear.** It is unclear when saved copies are set to be enveloped, so when users will see this framing is not known.\n\n**What**\n\nThe Read tool is how Claude opens and reads files. Its results can now be wrapped in framing text that marks where the file's content starts and ends. This happens only when the framing setting for a saved copy of the file is `enveloped`.\n\nIf the file itself contains text that looks like those marker tags, Claude Code neutralizes it so it cannot be mistaken for the real start or end. The connection Claude Code uses to talk to remote sessions also gained a marker recording when a whole file was held back.\n\n**Why**\n\nThis looks like protection against prompt injection, where text inside a file tries to pass itself off as instructions. The likely target is files synced from the cloud.\n\n- Area: Read Tool\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 3\/5\n- Scope: individual\n- Heads-up: no"}