Agent Ia Core

The workflow JSON

Copy or download the full n8n JSON below. Paste it into a new n8n workflow, add your credentials, activate. Full import guide →

{
  "name": "agent-ia-core",
  "nodes": [
    {
      "parameters": {
        "promptType": "define",
        "text": "={{ $json.text }}",
        "options": {
          "systemMessage": "=# \ud83e\udde0 Role  \nYou are an **AI agent** responsible for understanding requests received via WhatsApp and handling them using contextual memory and specialized tools connected through MCP.  \nToday\u2019s date is {{ $now }}.\n\n---\n\n# \ud83d\udccc General Capabilities  \nYou are able to:  \n- Understand and process incoming WhatsApp messages  \n- Use a **conversation memory** (20-message window) to maintain dialogue coherence  \n- Generate relevant responses using the **Gemini model**  \n- Use specialized tools to carry out user commands\n\n---\n\n# \ud83d\udd27 Available Tools (MCP)\n\n## MCP Gmail  \nAllows you to:  \n- Send an email  \n- Reply to an email  \n- Mark an email as read or unread  \n- Retrieve emails  \n\n**Use case:** Any request related to Gmail messaging.\n\n## MCP Calendar  \nAllows you to:  \n- Create an event  \n- Delete an event  \n- View an existing event  \n- Modify an existing event  \n\n**Use case:** Any request related to Google Calendar.\n\n---\n\n# \ud83d\udcec Gmail Tool Usage Guidelines\n\nYou have several Gmail tools depending on the email\u2019s read status:  \n- Use **Gmail - Unread Only** if the user requests only unread emails (e.g., \u201cWhat\u2019s my latest unread email?\u201d).  \n- Use **Gmail - All Mails** if the user asks to view emails without specifying read status.  \n- Use **Gmail - Read Only** if the user requests emails that have already been read.\n\n---\n\n# \ud83d\udccb Result Display (Emails or Events)\n\n- \ud83d\udfe1 If the request includes expressions like \u201cthe latest,\u201d \u201cmost recent,\u201d \u201ctoday\u2019s email,\u201d etc.  \n  \u2192 You must return **only one item**: the **most recent** or **most relevant**.\n\n- \ud83d\udfe2 Otherwise, if the user asks for a list or gives no specific constraint  \n  \u2192 Return **multiple items** in a **clear, concise, and structured** format.\n\n\u26a0\ufe0f Never return **multiple items** when the request clearly expects a single result.\n\n---\n\n# \ud83c\udd94 ID Handling\n\n- For every **email** or **event** you display, always **explicitly include the ID**.  \n  Example formats:  \n  ID: abc123 or Message_ID: xyz456\n\n- This ID must be **visible in your response**, even if the user didn\u2019t ask for it.\n\n- These IDs are essential for enabling **direct action** without further questions later.\n\n---\n\n# \ud83d\ude80 Direct Action Execution\n\nWhen a request includes a clear action (e.g., \u201creply,\u201d \u201cdelete,\u201d \u201cmark as read,\u201d etc.), **execute it immediately**.\n\n\u274c Never ask for confirmation before acting, unless the user expresses doubt or uses ambiguous wording.\n\n\u2705 If all required details (like the message or event ID) are available, act without asking.  \nIf the ID is missing, extract it from the relevant item or from memory if possible.\n\n---\n\n# \ud83d\uddc2\ufe0f Response Format  \n- Keep responses **clear and precise**  \n- Use **lists** when needed  \n- Ensure data is **well-organized** for easy reading\n\n---\n\n# \u26a0\ufe0f Constraints  \n- Never request **sensitive data** (passwords, personal or confidential information)  \n- Default response language is **English**, unless otherwise specified  \n- Never **hallucinate**: if any information is missing or unclear, **ask the user to clarify** rather than guessing  \n- If in doubt, **ask for confirmation** instead of risking an incorrect response\n"
        }
      },
      "type": "@n8n/n8n-nodes-langchain.agent",
      "typeVersion": 1.8,
      "position": [
        860,
        240
      ],
      "id": "d90024b3-d0c7-4d62-beb4-031153453369",
      "name": "AI Agent"
    },
    {
      "parameters": {
        "sseEndpoint": "https://your-instance.n8n.cloud/mcp/REDACTED_PATH/sse"
      },
      "type": "@n8n/n8n-nodes-langchain.mcpClientTool",
      "typeVersion": 1,
      "position": [
        1100,
        480
      ],
      "id": "35012c50-bcce-4f13-bf20-86858e4165c6",
      "name": "MCP Gmail"
    },
    {
      "parameters": {
        "sseEndpoint": "https://your-instance.n8n.cloud/mcp/REDACTED_PATH/sse"
      },
      "type": "@n8n/n8n-nodes-langchain.mcpClientTool",
      "typeVersion": 1,
      "position": [
        1220,
        480
      ],
      "id": "64824bc4-4c9b-4a49-b674-5f2a4a73d6ed",
      "name": "MCP Calendar"
    },
    {
      "parameters": {
        "updates": [
          "messages"
        ],
        "options": {}
      },
      "type": "n8n-nodes-base.whatsAppTrigger",
      "typeVersion": 1,
      "position": [
        460,
        240
      ],
      "id": "3ea97513-a574-4fe9-ac67-2071761ead0a",
      "name": "WhatsApp Trigger"
    },
    {
      "parameters": {
        "operation": "send",
        "phoneNumberId": "REDACTED_PHONE_NUMBER_ID",
        "recipientPhoneNumber": "REDACTED_PHONE",
        "textBody": "={{ $json.output }}",
        "additionalFields": {
          "previewUrl": false
        }
      },
      "type": "n8n-nodes-base.whatsApp",
      "typeVersion": 1,
      "position": [
        1220,
        240
      ],
      "id": "b6c6ad1a-9b4f-46df-b08b-770ebebb0f7b",
      "name": "WhatsApp Business Cloud"
    },
    {
      "parameters": {
        "assignments": {
          "assignments": [
            {
              "id": "REDACTED_ID",
              "name": "text",
              "value": "={{ $('WhatsApp Trigger').item.json.messages[0].text.body }}",
              "type": "string"
            },
            {
              "id": "REDACTED_ID",
              "name": "sessionId",
              "value": "={{ $('WhatsApp Trigger').item.json.contacts[0].wa_id }}",
              "type": "string"
            }
          ]
        },
        "options": {}
      },
      "type": "n8n-nodes-base.set",
      "typeVersion": 3.4,
      "position": [
        660,
        240
      ],
      "id": "bd79fa2e-6de8-4be1-b04b-2647156de390",
      "name": "Edit Fields"
    },
    {
      "parameters": {
        "modelName": "REDACTED_MODEL",
        "options": {}
      },
      "type": "@n8n/n8n-nodes-langchain.lmChatGoogleGemini",
      "typeVersion": 1,
      "position": [
        720,
        480
      ],
      "id": "4b68264f-457d-4e2f-958a-4cd37cd10db1",
      "name": "Google Gemini Chat Model"
    },
    {
      "parameters": {
        "sessionIdType": "customKey",
        "sessionKey": "={{ $json.sessionId }}",
        "contextWindowLength": 20
      },
      "type": "@n8n/n8n-nodes-langchain.memoryBufferWindow",
      "typeVersion": 1.3,
      "position": [
        880,
        480
      ],
      "id": "44ef02e6-8b02-4968-9998-c63cafdad475",
      "name": "Simple Memory"
    },
    {
      "parameters": {
        "content": "## \ud83e\udd16 AI Agent\n**Core of the workflow. This agent :**  \n- Interprets WhatsApp messages  \n- Uses Gemini to generate responses  \n- Maintains 20-message memory context  \n- Routes actions via MCP client tools (Gmail & Calendar)  \n\n\nPrompt includes full system instructions and tool definitions.  \nYou can modify the prompt inside the node under *System Message*.  \n\n\n**Connects:**  \n- Memory node \u2192 for session context  \n- Gemini model \u2192 for LLM generation  \n- WhatsApp input/output  \n- MCP client tools via tool connections  \n",
        "height": 620,
        "width": 1420
      },
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        0,
        0
      ],
      "typeVersion": 1,
      "id": "f1e1187f-c6c9-4634-b084-2da71174f850",
      "name": "Sticky Note1"
    },
    {
      "parameters": {},
      "type": "@n8n/n8n-nodes-langchain.toolThink",
      "typeVersion": 1,
      "position": [
        1320,
        480
      ],
      "id": "7c89a19f-2f66-4f18-ad94-35726f78d220",
      "name": "Think"
    }
  ],
  "connections": {
    "AI Agent": {
      "main": [
        [
          {
            "node": "WhatsApp Business Cloud",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "MCP Gmail": {
      "ai_tool": [
        [
          {
            "node": "AI Agent",
            "type": "ai_tool",
            "index": 0
          }
        ]
      ]
    },
    "MCP Calendar": {
      "ai_tool": [
        [
          {
            "node": "AI Agent",
            "type": "ai_tool",
            "index": 0
          }
        ]
      ]
    },
    "WhatsApp Trigger": {
      "main": [
        [
          {
            "node": "Edit Fields",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Edit Fields": {
      "main": [
        [
          {
            "node": "AI Agent",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Google Gemini Chat Model": {
      "ai_languageModel": [
        [
          {
            "node": "AI Agent",
            "type": "ai_languageModel",
            "index": 0
          }
        ]
      ]
    },
    "Simple Memory": {
      "ai_memory": [
        [
          {
            "node": "AI Agent",
            "type": "ai_memory",
            "index": 0
          }
        ]
      ]
    },
    "Think": {
      "ai_tool": [
        [
          {
            "node": "AI Agent",
            "type": "ai_tool",
            "index": 0
          }
        ]
      ]
    }
  },
  "active": false,
  "settings": {
    "executionOrder": "v1"
  },
  "versionId": "a3497b56-7af8-4914-abaa-f4e169760a31",
  "id": "vgmr1nM6k1IATPfP",
  "tags": []
}

About this workflow

agent-ia-core. Uses agent, mcpClientTool, whatsAppTrigger, whatsApp. Event-driven trigger; 10 nodes.

Source: https://github.com/AdelMessaoudi-13/WhatsApp-AI-Agent/blob/8761aa7ae38f9c4bdd24ed6416a3acf13798f794/workflows/agent_ia_core.json — original creator credit. Request a take-down →

More AI & RAG workflows → · Browse all categories →

Related workflows

Workflows that share integrations, category, or trigger type with this one. All free to copy and import.