{"version":"2.1.293","anchor":"new-shell-command-guidance-snippet-about-the-users-machine","canonical_anchor":"new-shell-command-guidance-snippet-about-the-users-machine","heading":"Claude gets guidance on writing commands your machine can check line by line","tier":"notice","area":"Bash","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.293\/e\/new-shell-command-guidance-snippet-about-the-users-machine","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.293","markdown":"### Claude gets guidance on writing commands your machine can check line by line\n\nNew instructions tell Claude to spell out paths and names literally in commands, since your machine may question lines it cannot fully read\n\n**Unclear.** It is not clear which sessions include these instructions, possibly only remote or cloud ones.\n\n**What**\n\nClaude Code has new instructions for the model about commands it writes to run on your machine. They say:\n\n- Your machine checks each command line before running it, and may ask you about, or refuse, a line it cannot fully read.\n\n- Paths, names and ids should be written out literally rather than built from shell variables, `for` loops or `$(\u2026)`.\n\n- Chaining commands with `&&` or `;` is fine.\n\n- The `$(cat <<'EOF'` form for commit messages is fine.\n\n**Why**\n\nThis should change how Claude writes shell commands, so fewer of them trigger a question or a refusal from your machine.\n\n- Area: Bash\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 3\/5\n- Scope: individual\n- Heads-up: no"}