Why the CLI
An interface agents already know
For integration work that fits the command line, the CLI is faster to reason about and cheaper to run than a full MCP tool surface.
- 01
A CLI-native interface
Agents use commands, flags, and help text they already understand, with no new tool schema to learn before they can act.
- 02
Fewer tokens
No large tool schema sits in context. Agents call commands directly and look up help only when they need it.
- 03
Easy setup
Install, sign in, and connect a provider in minutes, then run the same workflow locally, in scripts, or in CI.
Built for production
CLI speed, Metorial control
Raw CLI access is fast, but it is not enough for production agents. Metorial keeps CLI-driven integration work governed by identity, access policies, tracing, and Protoguard before anything reaches your systems.
Identity and access built in
CLI workflows stay tied to the same people, groups, agents, providers, and tools you govern across Metorial.
- Sign in through SSO and SAML
- Use tokenless auth in scripts and CI
- Apply access policiesby user, group, agent, provider, and tool
Every action is traceable
Metorial records who asked, which agent acted, what command ran, and how each provider request was handled.
- Review every call in Tracing
- Keep accountability with audit logs
- Monitor CLI-driven work through observability
Protected before it reaches systems
Protoguard reviews risky requests and policy decisions before CLI-driven agent work touches connected integrations.
- Block unsafe requests with Protoguard
- Detect prompt injection before an agent acts through monitoring
- Catch risky provider changes with schema monitoring
Frequently asked questions
Common questions about CLI.
What is the CLI?
The CLI is the command-line interface for Metorial. It lets developers and AI agents connect providers, run integration workflows, test tools, and deploy governed access from the terminal, alongside the Metorial API and official SDKs.How do I install the Metorial CLI?
The CLI is published on the public npm registry as @metorial/cli. Install it globally with "npm install -g @metorial/cli", then run "metorial login" to authenticate. Where npm is unavailable, pipe the shell installer at metorial.com/install.sh into sh. The package page is at npmjs.com/package/@metorial/cli, and the full command reference is at cli.md.Why use a CLI instead of MCP for agents?
For command-shaped workflows, a CLI can keep agents close to interfaces they already understand: commands, flags, help text, scripts, and readable output. Metorial adds the enterprise controls raw CLI access usually lacks, including access policies, tracing, and tokenless auth.Does Metorial still support MCP?
Yes. Use MCP when you want protocol-native integrations, dynamic tool discovery, or custom servers. Use the CLI when the workflow is better expressed as commands and automation. Both can live under the same Metorial governance model through custom MCP, providers, and policies.Are CLI-driven agent actions traced?
Yes. Metorial records the agent, user, provider, tool, request, response, deploy, and policy decision behind CLI-driven activity, so security and platform teams can review what happened through audit logs and observability.How does access control work with CLI workflows?
CLI workflows follow the same policies and access control as the rest of Metorial. You can scope access by user, group, agent, provider, and tool, then block actions that fall outside policy before they reach connected systems.Can teams use SSO or SAML with the CLI?
Yes. Teams can sign in with existing SSO and SAML, and the CLI uses tokenless auth so scripts and CI do not need long-lived provider credentials.How does Protoguard protect CLI-driven integrations?
Protoguard reviews agent messages and tool requests, watches for risky provider changes, and blocks anything outside your policies before it reaches the systems your CLI workflows connect to.