Server Dashboard
The default tab when opening any server. A real-time command center for monitoring performance, watching tool executions as they happen, and managing server configuration — all from a single view.
Most MCP platforms give you a list of servers and nothing else. Vinkius Cloud provides per-server observability at the level you would expect from a production infrastructure platform.
Server banner
The banner at the top of every server detail view shows identity, status, and governance at a glance:
Banner actions
| Action | Description |
|---|---|
| Quick Connect | Generates a ready-to-paste MCP configuration for AI clients |
| Sync | Re-imports tools from the upstream OpenAPI spec (OpenAPI servers only) |
| Edit | Opens the server configuration modal for per-server overrides |
| Delete | Permanently removes the server, tokens, and deployment history |
Status badges
| Badge | Description |
|---|---|
| ONLINE / OFFLINE | Clickable toggle — disable a server instantly without deleting it |
| N tools | Total registered MCP tools |
| N groups | Number of tag groups (grouped mode) or FLAT indicator |
| DLP | Data Loss Prevention is active |
| FinOps | FinOps Guard is active |
Infrastructure stats
A 4×2 real-time metrics grid, identical in structure to the fleet overview but scoped to the individual server:
| Metric | Insight |
|---|---|
| Requests 24h | Total tool calls. Spot unexpected spikes or drops instantly. |
| P95 Latency | Worst-case latency for 95% of requests. Critical for SLA compliance. |
| Error Rate | Color-coded: green at 0%, amber at moderate, red above 5%. |
| Avg Response | Baseline for performance regression detection. |
| Slowest Tool | Identifies the bottleneck tool by average response time. |
| Busiest Tool | Identifies the most-called tool — useful for capacity planning. |
| Active Clients | Currently connected SSE sessions — see who is consuming your server. |
| Last Activity | Time since the last request — quickly detect stale or abandoned servers. |
Stats refresh automatically when you return to the Dashboard tab.
Live request feed
Below the stats grid, a real-time streaming feed displays tool executions as they happen. Each entry shows:
The feed columns:
| Column | Description |
|---|---|
| Timestamp | When the tool was called |
| Tool Name | Which MCP tool was executed |
| Semantic Verb | QUERY (read), MUTATION (write), DESTRUCTIVE (delete) — color-coded red for destructive operations |
| Latency | Response time in milliseconds |
| Status Code | HTTP response: green (2xx), amber (4xx), red (5xx) |
The streaming indicator (green pulse dot) confirms data is updating in real time. The feed shows the 10 most recent executions and refreshes on every tab switch.
Marketplace privacy guard
When a server has an active Marketplace listing, the live feed is replaced with a privacy notice. This prevents sellers from seeing individual buyer request data — request details are never loaded into the browser, not even accessible via DevTools.
Quick Connect
The one-click integration point. Quick Connect generates a ready-to-paste configuration for any MCP-compatible AI client:
{
"mcpServers": {
"my-api-server": {
"url": "https://edge.vinkius.com/{token}/mcp"
}
}
}Paste this into Claude Desktop, Cursor, Windsurf, VS Code with Copilot, or any client that supports the MCP protocol. The URL includes your connection token for authenticated access.
The Quick Connect modal provides:
- SSE URL — The complete endpoint URL
- JSON Config — Copy-paste configuration block
- Client selector — See integration instructions for popular AI clients
TIP
Quick Connect is also available from the token creation modal. Generate a token and immediately get the configuration — no extra steps required.
Server configuration
The Edit modal opens inline from the dashboard. This is where per-server governance overrides are applied:
Configurable settings
| Setting | Global Default | Per-Server Override |
|---|---|---|
| Name | Set at creation | Editable |
| Description | Optional | Editable |
| Base URL | Set at creation | Editable (except Agent Skills) |
| DLP Protection | Inherits from Settings → Data Shielding | Toggle + custom patterns per server |
| DLP Patterns | Inherits global patterns | Add or replace patterns per server |
| FinOps Guard | Inherits from Settings → FinOps Guard | Toggle per server |
| Max Payload Size | Inherits global limit | Custom slider per server |
| Max Payload Bytes | Inherits global OOM Shield | Custom limit per server |
| Tool Exposition | From MCP Defaults | Flat, Grouped, or None |
Inheritance model
Settings follow a global-to-per-server chain. When a per-server override is not set, the global default from Settings applies. This means you can configure governance once and have it apply to every server — then selectively override where needed.
Server lifecycle actions
Beyond editing, the dashboard provides critical lifecycle controls:
Status toggle
Click the ONLINE / OFFLINE badge to instantly enable or disable a server. When disabled:
- The server stops accepting MCP requests
- Existing SSE connections are terminated
- The server remains in your account and can be re-enabled anytime
- All tokens, tools, and configuration are preserved
Kill all connections
Force-close all active SSE connections immediately. Useful for emergencies or maintenance windows where you need to guarantee zero active sessions before making changes.
Sync tools
For OpenAPI-based servers, the Sync button re-reads the upstream OpenAPI specification and updates the tool list. New endpoints become new tools, removed endpoints are deleted, and modified endpoints are updated.
The sync process runs asynchronously with a 3-second polling delay to account for large API specs.
Delete server
Permanently and irreversibly removes the server. If the server has active Marketplace subscribers, the platform warns with the active subscriber count and last billing cycle end date before allowing deletion.
For servers with subscribers, a force-scheduled deletion waits until all current billing cycles end before removing the server.
Next steps
Frequently Asked Questions
What information does the MCP server dashboard show?
The dashboard provides a real-time command center with eight infrastructure metrics (requests, P95 latency, error rate, avg response, slowest tool, busiest tool, active clients, last activity), a live streaming feed of tool executions, Quick Connect integration, and full server configuration management.
Does the dashboard update in real time?
Yes. The infrastructure stats refresh automatically when you return to the Dashboard tab. The live request feed streams tool executions as they happen via SSE (Server-Sent Events), with a green pulse indicator confirming real-time data flow.
Can I configure DLP and FinOps Guard per server from the dashboard?
Yes. Click the Edit button to open the server configuration modal. You can toggle DLP Protection and FinOps Guard per server, add custom DLP patterns, set max payload size and bytes limits, and configure tool exposition mode — all without leaving the dashboard.
What is the Quick Connect feature?
Quick Connect generates a ready-to-paste MCP configuration for any AI client. It provides the SSE URL, JSON config block, and client-specific instructions for Claude Desktop, Cursor, VS Code with Copilot, Windsurf, and Antigravity. One click to copy, paste into your client, done.
Can I disable a server without deleting it?
Yes. Click the ONLINE/OFFLINE badge to instantly toggle server status. When disabled, the server stops accepting MCP requests and existing SSE connections are terminated, but all configuration, tokens, tools, and deployment history are preserved.
What happens when I delete a server with Marketplace subscribers?
The platform warns with the active subscriber count and last billing cycle end date. A force-scheduled deletion waits until all current billing cycles end before removing the server. This protects buyer subscriptions from abrupt service termination.