Quickstart#
Get FirstOps running in your environment in under 5 minutes.
Prerequisites#
- A FirstOps account (sign up)
- The FirstOps CLI installed on your machine
- At least one MCP-compatible AI tool (Cursor, Claude Code, VS Code, etc.)
Step 1: Install the CLI#
curl -fsSL https://firstops-cli-releases.s3.amazonaws.com/install.sh | sh
Step 2: Authenticate#
fo auth login
This opens your browser for authentication. Once complete, you're authenticated and ready to go. (See Identity & Auth for how the cryptographic authentication works under the hood.)
Step 3: Run setup#
fo setup
Setup does three things:
- Starts the daemon — a background process that monitors MCP configurations
- Scans for connections — discovers all MCP server configs across your AI tools
- Reports discoveries — unmonitored connections appear in your dashboard
Step 4: View your connections#
Open the FirstOps dashboard to see discovered connections. Each connection shows:
| Field | Description |
|---|---|
| Tool | The upstream service (Notion, GitHub, Slack, etc.) |
| Client | The AI tool using it (Cursor, Claude Code, etc.) |
| Status | monitored or unmonitored |
| Principal | Who owns this connection |
Step 5: Fix unmonitored connections#
Fix connections from the CLI:
fo discover fix <connection-id>
Or fix directly from the FirstOps dashboard — select an unmonitored connection and click Fix. The daemon picks up the change on its next sync.
Either way, this rewrites the local MCP configuration to route through the FirstOps gateway, enabling policy enforcement and audit logging.
What happens next?#
Once connections are monitored:
- All requests flow through the FirstOps proxy
- Policies are evaluated on every request
- Credentials are brokered (agents never see raw tokens)
- Every interaction is logged in the audit trail
Your admin can configure policies and access groups to control what agents can do.