Global flags
Flags accepted by every command in the unified Prisma CLI.
All commands in the unified Prisma CLI accept these flags.
| Flag | What it does |
|---|---|
--json | Print machine-readable output (shorthand for --format json). Use this in CI and scripts. |
--format <mode> | Output format: human, json, or markdown. |
--log-level <level> | Commentary verbosity: error, warn, info, or verbose. |
-q, --quiet | Suppress nonessential output (shorthand for --log-level error). |
-v, --verbose | Print more detail (shorthand for --log-level verbose). |
--color / --no-color | Force colored output on or off. |
--interactive / --no-interactive | Force prompts on or off. |
-y, --yes | Accept prompt defaults without asking. |
--confirm <token> | Grant a consent prompt non-interactively by typing its token (repeatable). |
--config <path> | Read this config file instead of ./prisma.config.ts. |
-h, --help | Print the manual for a command: what it does, its options, a workflow where one applies, and examples. |
--version | Print the CLI version and exit. |
Output formats
Without --format, a terminal gets human output and a pipe gets json. markdown is never chosen for you; pass --format markdown to get it.
markdown is for AI agents that read command output as text rather than parsing JSON. It prints the same content as human, as plain Markdown: a summary line, label: value lines, pipe tables, bullet lists, and fenced code, followed by a ### Next section with the suggested next commands and a ### Diagnostics section with any findings. There is no padding, alignment, or color, and no JSON envelope. Everything the command produces, including errors, help, --version, and progress, goes to stdout, so an agent reads one stream.
Use npx prisma@latest <command> --help when you need the exact command help from the installed version.
The output modes and the JSON envelope these flags select are documented on Configuration.