# CLI reference One package, `@morigraph/cli`, with one command, `morigraph`. It needs Node.js 22 or newer and has no dependencies. ```sh npm install -g @morigraph/cli # or run any command as: npx @morigraph/cli@latest ``` ## Sign in ```sh morigraph login ``` Prints a code and a link, opens the link if a person is at the terminal, and waits for a workspace admin to approve. The key is stored in `~/.config/morigraph/credentials.json`, readable only by you. See [Authentication](https://morigraph.com/docs/auth.md). | Option | Effect | |---|---| | `--no-browser` | Print the link without opening it | | `--with-key` | Read an existing key from standard input and store it: `printf %s "$KEY" \| morigraph login --with-key` | | `--name NAME` | The name the new key gets in the dashboard | `MORIGRAPH_API_KEY`, when set, is used instead of the stored key. `morigraph logout` forgets the stored key on this machine; it does not revoke the key. ## Look around ```sh morigraph whoami # workspace, key and credits available morigraph credits # balance, and this key's last 30 days morigraph tools # every tool, its price, and whether it is switched on morigraph tools search image # filter by text morigraph tools schema image_gen # one tool's arguments as JSON Schema ``` ## Call a tool ```sh morigraph call web_search --arg query="rate limiting patterns" --arg num_results=3 morigraph call image_gen --json-args '{"prompt": "a lighthouse at dusk", "size": "landscape"}' echo '{"query": "rate limiting patterns"}' | morigraph call web_search --stdin --yes ``` | Option | Effect | |---|---| | `--arg key=value` | One argument. Numbers, `true`, `false`, and JSON arrays or objects are parsed; anything else is a string. Repeat for more | | `--json-args JSON` | All arguments as one JSON object | | `--stdin` | Read the JSON object from standard input | | `--yes` | Spend credits without asking. Required when no person is at the terminal. `MORIGRAPH_YES=1` does the same | | `--wait` | For a tool that returns a job, wait until it finishes | | `--idempotency-key KEY` | Reuse a key to retry a call safely. One is generated when you give none | At a terminal, `call` shows the price and asks before spending. ## Jobs ```sh morigraph jobs get JOB_ID # status now morigraph jobs wait JOB_ID # wait until it finishes ``` ## Set up a coding agent ```sh morigraph setup agent ``` Installs a small skill that teaches Claude Code and Codex the commands on this page, for whichever of the two is present on the machine. It prints a line to paste to the agent. A skill you have edited is never overwritten. The agent still needs a key: run `morigraph login` or set `MORIGRAPH_API_KEY`. ## Connect an assistant over MCP `morigraph setup --client claude-code` (or `codex`, `claude`) is a different job: it connects an assistant to your workspace's files, memory, skills and connected apps over MCP, with OAuth and no API key. See [Connect](https://morigraph.com/connect.md). ## Output and exit codes Output is JSON when `--json` is given or when standard output is not a terminal, so scripts and agents always get one JSON object per line. At a terminal it is plain text. | Exit code | Meaning | |---|---| | 0 | Success | | 1 | Error. In JSON mode the output has `status: "error"`, `code` and `message` | | 2 | Sign-in needed, or a manual step remains | | 3 | Out of credits | ## Safety - A key is read from `MORIGRAPH_API_KEY` or the stored login, never from the command line, and is never printed. - A key is sent only to morigraph. There is no option to change the address. - No command waits for input unless a person is at the terminal. - No telemetry and no install scripts. ## Built-in help This is the output of `morigraph --help` for the current release: ```text morigraph CLI Package: @morigraph/cli · executable: morigraph · legacy alias: brainlet Connect an assistant (MCP, signs in with OAuth, no API key): morigraph setup --client claude-code|codex|claude|manual [options] morigraph doctor --client claude-code|codex [options] Call tools from a terminal, script or agent (uses an API key): morigraph login [--env prod|dev] [--no-browser] Sign in; a workspace admin approves in the browser. morigraph login --with-key Read a key from standard input and store it. morigraph logout Forget the stored key on this machine. morigraph whoami The workspace, key and credit balance. morigraph tools [list | search TEXT | schema NAME] morigraph call NAME [--arg key=value ...] [--json-args JSON | --stdin] [--file PATH] [--wait] [--yes] morigraph jobs get|wait JOB_ID morigraph credits morigraph setup agent [--client claude-code|codex] Install the skill that teaches an agent these commands. Setup options: --env prod|dev Production by default. Fresh dev uses morigraph-dev. --server-name NAME Use an explicit MCP server name without replacing another. --dry-run Show the plan; no commands, login, or writes. --no-login Configure the host and skill; print the login step. Everywhere: --json Machine-readable output. The default when output is not a terminal. --yes Allow a call to spend credits without asking (or set MORIGRAPH_YES=1). --version Show the CLI version. --help Show this help. API keys come from MORIGRAPH_API_KEY or `morigraph login`, never from the command line, and are sent only to morigraph. Exit codes: 0 ok, 1 error, 2 sign-in or a manual step needed, 3 out of credits. Docs: https://morigraph.com/docs/ For AI agents: https://morigraph.com/docs/agents.md Setup runs on the machine that configures your host. Automated setup supports macOS and Linux. Claude chat and Windows receive manual connection instructions. OAuth credentials stay with the host. No telemetry, remote scripts, custom URLs or installation lifecycle hooks are used. Matching existing brainlet/brainlet-dev connections are reused automatically. ``` --- Canonical page: https://morigraph.com/docs/cli/ Updated: 2026-11-11 All developer docs: https://morigraph.com/docs.md · For AI agents: https://morigraph.com/docs/agents.md · OpenAPI: https://api.morigraph.com/openapi.json · Site overview: https://morigraph.com/llms.txt