Dashboard Explorer

Interactive simulation of the MCP Orchestrator's observability dashboards

About This Demo: This interactive simulation shows how the MCP Orchestrator provides full observability through 4 Grafana dashboards, 15+ Prometheus metrics, and REST API endpoints. Click the tabs below to explore different dashboard views.

Live Simulation
MCP Overview
M5 Learning Layer
API Endpoints
Prometheus Metrics
0
Total Requests
0ms
Avg Latency
100%
Success Rate
6
Active Tools

Orchestrator

Core routing engine healthy

Dashboard API

FastAPI server running

Prometheus

Metrics collection active

Grafana

4 dashboards provisioned

Request Rate (Last Hour)

Learning Layer

Overall health status

Vector Store (LanceDB)

Semantic similarity engine

Graph Store (Neo4j)

Relationship mapping

0
Total Embeddings
0
Task Nodes
6
Tool Nodes
0
Relationships

Embeddings by Tool

claude_code_gen (60%)
claude_general (25%)
internal_calc (15%)

Embeddings by Outcome

Success (85%)
Failure (15%)

Available Endpoints

Endpoint Method Description Calls
/health GET System health check 0
/tools GET List available tools 0
/execute POST Execute a task 0
/config GET View configuration 0
/learning/stats GET Learning layer statistics 0
/learning/health GET Learning layer health 0
/events/stream GET SSE event stream 0
/logs/stream GET SSE log stream 0

Sample API Response: /learning/stats

{
  "healthy": true,
  "vector_store": {
    "available": true,
    "total_embeddings": 0,
    "embeddings_by_tool": {},
    "embeddings_by_outcome": {}
  },
  "graph_store": {
    "available": true,
    "task_count": 0,
    "tool_count": 6,
    "capability_count": 12,
    "outcome_count": 2,
    "relationship_count": 0
  }
}

Prometheus Metrics

15+ custom metrics exposed at /metrics/prometheus

Metric Name Type Value Description
mcp_learning_healthy Gauge 1 Overall learning layer health
mcp_learning_vector_store_available Gauge 1 LanceDB availability
mcp_learning_graph_store_available Gauge 1 Neo4j availability
mcp_learning_vector_embeddings_total Gauge 0 Total embeddings in LanceDB
mcp_learning_graph_tasks_total Gauge 0 Task nodes in Neo4j
mcp_learning_graph_tools_total Gauge 6 Tool nodes in Neo4j
mcp_learning_graph_relationships_total Gauge 0 Relationships in Neo4j
mcp_requests_total Counter 0 Total HTTP requests
mcp_request_latency_ms Histogram - Request latency distribution

Grafana Dashboards

4
Dashboards
15+
Metrics
2
Datasources
Dashboard Purpose
MCP Overview System health, request rates, latency metrics
M5 Learning Layer LanceDB embeddings, Neo4j nodes, relationships
API Endpoints Per-endpoint request metrics and latency
Logs Explorer Full-text log search via Loki