Masheev CLI
Release status. The
masheevCLI is published on npm as@masheev/cli. Runmasheev --helpto see the commands in your installed version.
npm install -g @masheev/cli # Node.js 20+
masheev upgrade # later: update to the latest release
masheev login # browser sign-in
masheev login --send # also allow sending customer replies
masheev login --token < key # agent key from Developers → Connect an agent
masheev whoami
masheev login opens your browser and waits for the redirect on http://127.0.0.1:<port>/callback (authorization code with PKCE). On the consent page, pick one workspace, how long access lasts and which write operations to allow. Sending to customers is a separate switch, off by default. Masheev may grant less than the CLI asked for; masheev whoami shows what you have. Use --no-browser to print the URL instead. On a machine without a browser, sign in with an agent key (--token); one-time-code sign-in (--device) is pending server support.
Credentials go in your OS keychain. In CI, set MASHEEV_API_TOKEN to an agent key instead. Use --profile or --org when you have several workspaces.
Run any operation
masheev ops list
masheev ops describe contacts.update
masheev contacts list --limit 20 --json
masheev ops run contacts.update --input-json @patch.json
Commands are generated from the operation catalog your credential can use; masheev ops describe shows each operation's effect and required permissions. Operations that delete or send ask for confirmation, or need --yes in scripts. Billing, irreversible deletes and credentials are unavailable, as in every Masheev agent tier.
Inbox, replies and agents
masheev inbox --status open
masheev inbox show <conversation>
masheev inbox assign <conversation> --to me
masheev reply <conversation> --message "Thanks, booked for 3pm"
masheev reply confirm <draft-id> --yes
masheev agents pause <agent-id>
Replies are two steps: the server creates a draft for the conversation's existing contact, and nothing is sent until you confirm it. Sending needs masheev login --send.
Connect AI hosts
masheev mcp install claude # preview the change
masheev mcp install claude --write # apply it (old file backed up)
masheev mcp install codex|cursor|openclaw|hermes --write
masheev mcp serve # local stdio server; replaces @masheev/mcp
masheev skills install
mcp install configures the hosted server with OAuth and writes no secret; the client then signs in.
Scripting
--jsonprints one JSON value on stdout; errors are JSON on stderr.- Exit codes: 0 ok, 1 error, 2 usage, 3 not signed in, 4 forbidden, 5 not found, 6 conflict, 7 rate limited, 8 unavailable, 9 outcome unknown (check before retrying), 10 not supported yet, 11 confirmation required.
- A missing permission names the login that grants it, such as
masheev login --send. A rate limit reports how long to wait. A 503 can mean the workspace's agent access is paused; don't replay changes automatically. - Prompts only appear in an interactive terminal.
NO_COLORdisables colour.