pub const HELP: &str = "Basalt — embedded SQL database\n\n\
Usage:\n basalt [OPTIONS] [DATABASE_PATH | :memory:]\n\n\
Options:\n -c, --command SQL Execute SQL and exit; may be repeated\n -f, --file PATH Execute a SQL script and exit; '-' reads stdin\n -o, --output FORMAT Result format: table, csv, or json\n --table Use table output (the default)\n --csv Use CSV output\n --json Use JSON-lines output\n --no-header Omit column headers in table/CSV output\n --quiet Suppress non-query success messages\n -h, --help Print this help\n -V, --version Print the version\n\n\
Interactive commands:\n .help Show this help\n .tables List tables\n .schema [TABLE] Show CREATE TABLE statements\n .mode table|csv|json Change result format\n .headers on|off Toggle result headers\n .checkpoint Flush the snapshot and truncate the WAL\n .show Show frontend state\n .clear Discard the pending SQL buffer\n .quit, .exit Leave the shell\n\n\
MCP server:\n basalt mcp [OPTIONS] [DATABASE_PATH | :memory:]\n\n\
Workspace:\n basalt init PATH\n basalt workspace --help\n\n\
JSON output is one JSON object per statement (JSON Lines). CSV output emits\n\
only query rows, so it can be piped directly into another data tool. Each CLI\n\
SQL action and interactive input buffer is limited to 16 MiB.\n";