| 1 | # AI Chat with Netdata |
| 2 | |
| 3 | Chat with your infrastructure using natural language through two distinct integration architectures. |
| 4 | |
| 5 | ## Integration Architecture |
| 6 | |
| 7 | ### Method 1: Connect to a Local Agent or Parent |
| 8 | |
| 9 | ```mermaid |
| 10 | flowchart TB |
| 11 | LLM("LLM Provider<br/>OpenAI, Anthropic, etc.") |
| 12 | |
| 13 | subgraph infra["Your Infrastructure"] |
| 14 | direction TB |
| 15 | subgraph userLayer[" "] |
| 16 | direction LR |
| 17 | User("User") |
| 18 | Client("AI Client<br/>Claude Desktop, Cursor, etc.") |
| 19 | |
| 20 | User -->|"(1) Ask question"| Client |
| 21 | Client -->|"(8) Display response"| User |
| 22 | end |
| 23 | |
| 24 | Agent("Netdata Agent or Parent<br/>with MCP Server") |
| 25 | |
| 26 | Client -->|"(4) Execute tools"| Agent |
| 27 | Agent -->|"(5) Return data"| Client |
| 28 | end |
| 29 | |
| 30 | Client -->|"(2) Send query"| LLM |
| 31 | LLM -->|"(3) Tool commands"| Client |
| 32 | Client -->|"(6) Send results"| LLM |
| 33 | LLM -->|"(7) Final answer"| Client |
| 34 | |
| 35 | %% Style definitions |
| 36 | classDef alert fill:#ffeb3b,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px |
| 37 | classDef neutral fill:#f9f9f9,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px |
| 38 | classDef complete fill:#4caf50,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px |
| 39 | classDef database fill:#2196F3,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px |
| 40 | |
| 41 | %% Apply styles |
| 42 | class User alert |
| 43 | class Client neutral |
| 44 | class LLM complete |
| 45 | class Agent complete |
| 46 | class infra,userLayer database |
| 47 | ``` |
| 48 | |
| 49 | **How it works:** |
| 50 | |
| 51 | 1. You ask a question to your AI client |
| 52 | 2. LLM responds with tool execution commands |
| 53 | 3. Your AI client executes tools against Netdata Agent MCP (locally) |
| 54 | 4. Your AI client sends tool responses back to LLM |
| 55 | 5. LLM provides the final answer |
| 56 | |
| 57 | **Key characteristics:** |
| 58 | |
| 59 | - Your AI client orchestrates all communication |
| 60 | - Netdata Agent MCP runs locally on your infrastructure |
| 61 | - No internet access required for Netdata Agent |
| 62 | - Full control over data flow and privacy |
| 63 | |
| 64 | ### Method 2: Connect to Netdata Cloud |
| 65 | |
| 66 | ```mermaid |
| 67 | flowchart TB |
| 68 | LLM("LLM Provider<br/>OpenAI, Anthropic, etc.") |
| 69 | CloudMCP("Netdata Cloud MCP<br/>app.netdata.cloud/api/v1/mcp") |
| 70 | |
| 71 | subgraph infra["Your Infrastructure"] |
| 72 | direction TB |
| 73 | subgraph userLayer[" "] |
| 74 | direction LR |
| 75 | User("User") |
| 76 | Client("AI Client<br/>Claude Desktop, Cursor, etc.") |
| 77 | |
| 78 | User -->|"(1) Ask question"| Client |
| 79 | Client -->|"(8) Display response"| User |
| 80 | end |
| 81 | |
| 82 | Agents("Netdata Agents<br/>and Parents") |
| 83 | end |
| 84 | |
| 85 | Client -->|"(2) Send query"| LLM |
| 86 | LLM -->|"(3) Tool commands"| Client |
| 87 | Client -->|"(4) Execute tools"| CloudMCP |
| 88 | CloudMCP -->|"(5) Return data"| Client |
| 89 | Client -->|"(6) Send results"| LLM |
| 90 | LLM -->|"(7) Final answer"| Client |
| 91 | |
| 92 | Agents -.->|"Stream metrics"| CloudMCP |
| 93 | |
| 94 | %% Style definitions |
| 95 | classDef alert fill:#ffeb3b,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px |
| 96 | classDef neutral fill:#f9f9f9,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px |
| 97 | classDef complete fill:#4caf50,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px |
| 98 | classDef database fill:#2196F3,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px |
| 99 | |
| 100 | %% Apply styles |
| 101 | class User alert |
| 102 | class Client neutral |
| 103 | class LLM complete |
| 104 | class CloudMCP,Agents complete |
| 105 | class infra,userLayer database |
| 106 | ``` |
| 107 | |
| 108 | **How it works:** |
| 109 | |
| 110 | 1. You ask a question to your AI client |
| 111 | 2. Your AI client sends the query to the LLM |
| 112 | 3. LLM responds with tool execution commands |
| 113 | 4. Your AI client executes tools against |
| 114 | Netdata Cloud MCP (over the internet) |
| 115 | 5. Netdata Cloud MCP returns the data |
| 116 | 6. Your AI client sends tool responses back to LLM |
| 117 | 7. LLM provides the final answer |
| 118 | 8. Your AI client displays the response |
| 119 | |
| 120 | **Key characteristics:** |
| 121 | |
| 122 | - Same client-controlled architecture as Method 1, but with a cloud-hosted MCP endpoint |
| 123 | - No direct network access to Netdata Agents required |
| 124 | - Complete visibility across all nodes in your infrastructure |
| 125 | - Any MCP-aware client can connect |
| 126 | |
| 127 | ## Quick Comparison |
| 128 | |
| 129 | | Aspect | Method 1: Local Agent/Parent | Method 2: Netdata Cloud | |
| 130 | |--------|---------------------------|---------------------| |
| 131 | | **Setup Complexity** | Moderate (configure AI client + local MCP) | Simple (configure AI client + cloud endpoint) | |
| 132 | | **Data Privacy** | Depends on LLM provider | Depends on LLM provider | |
| 133 | | **Internet Requirements** | AI client needs internet, MCP is local | Both AI client and MCP need internet | |
| 134 | | **Supported AI Clients** | Any MCP-aware client | Any MCP-aware client | |
| 135 | | **Infrastructure Access** | One Agent or Parent's scope | All infrastructure | |
| 136 | | **Network Access** | Direct access to Netdata IP | No firewall changes | |
| 137 | |
| 138 | To get started with Method 2, see |
| 139 | [Netdata Cloud MCP](/docs/netdata-ai/mcp/README.md#netdata-cloud-mcp) |
| 140 | for setup instructions. |