Conversation Ai Create Agent

Immutable Catalog Entry

This descriptor comes from the active versioned Portal catalog.

WRITEghl.conversation_ai_create_agent

• Contract tier: agent_ready
• Risk: write
• Catalog version: 2026-07-12.2
• Descriptor hash: f3f465bcadf5287b7fb062c15ec0a1235f0c6972e500d16377e9457142878b57

Create a GHL Conversation AI agent. GHL category: AI Agents / Conversation AI. This writes to GHL Conversation AI agent/action configuration. Do not modify production agents during validation unless the user explicitly provides disposable fixtures.

Request Body

{
  "jsonrpc": "2.0",
  "id": "ghl-conversation-ai-create-agent-example",
  "method": "tools/call",
  "params": {
    "name": "ghl.conversation_ai_create_agent",
    "arguments": {
      "body": {},
      "query": {},
      "tenantId": "tenantId_123",
      "locationId": "locationId_123",
      "strictTenant": false,
      "includeFullConfig": false
    }
  }
}

Arguments Schema

{
  "type": "object",
  "required": [
    "body"
  ],
  "properties": {
    "body": {
      "type": "object",
      "required": [
        "name",
        "personality",
        "goal",
        "instructions"
      ],
      "properties": {
        "goal": {
          "type": "string",
          "description": "Agent goal/configuration text. Redacted from default MCP output."
        },
        "mode": {
          "type": "string",
          "description": "Optional agent mode when documented by GHL, such as off for a disabled/disposable agent."
        },
        "name": {
          "type": "string",
          "description": "Agent name. Use a clear disposable prefix during audit tests."
        },
        "personality": {
          "type": "string",
          "description": "Agent personality/configuration text. Redacted from default MCP output."
        },
        "instructions": {
          "type": "string",
          "description": "Agent instructions/configuration text. Redacted from default MCP output."
        },
        "knowledgeBaseIds": {
          "type": "array",
          "items": {
            "type": "string"
          },
          "description": "Optional knowledge base IDs. Redacted from default MCP output."
        }
      },
      "description": "Request body for a GHL Conversation AI agent. GHL requires name, personality, goal, and instructions for create/update agent flows.",
      "additionalProperties": true
    },
    "query": {
      "type": "object",
      "description": "Optional GHL-supported query parameters for this official Conversation AI endpoint, such as pagination, agent, or generation filters from the official docs. Do not include locationId unless you only need local safety validation. For conversation_ai_search_agents, search/q/name are accepted but applied locally because GHL rejects those query fields.",
      "additionalProperties": true
    },
    "tenantId": {
      "type": "string",
      "description": "Tenant ID from list_tenants."
    },
    "locationId": {
      "type": "string",
      "description": "Optional GHL location ID for safety validation only. If provided, it must match the authenticated runtime location; Conversation AI endpoints are scoped by the PIT token and this field is not forwarded to GHL."
    },
    "strictTenant": {
      "type": "boolean",
      "default": true,
      "description": "Require tenantId to match locationId routing (default: true)."
    },
    "includeFullConfig": {
      "type": "boolean",
      "description": "Defaults to false. Set true only when the user explicitly needs full Conversation AI configuration such as personality, goal, instructions, prompts, knowledge-base IDs, or action parameters."
    }
  },
  "additionalProperties": false
}

Output Schema

{
  "type": "object",
  "required": [
    "ok",
    "data",
    "error",
    "meta"
  ],
  "properties": {
    "ok": {
      "type": "boolean",
      "description": "Whether the GHL MCP operation completed successfully."
    },
    "data": {
      "anyOf": [
        {
          "type": "object",
          "additionalProperties": true
        },
        {
          "type": "array",
          "items": {}
        },
        {
          "type": "string"
        },
        {
          "type": "number"
        },
        {
          "type": "boolean"
        },
        {
          "type": "null"
        }
      ],
      "description": "Tool-specific GHL or local navigation result when ok is true."
    },
    "meta": {
      "type": "object",
      "description": "Safe execution, routing, and timing metadata.",
      "additionalProperties": true
    },
    "error": {
      "anyOf": [
        {
          "type": "object",
          "additionalProperties": true
        },
        {
          "type": "null"
        }
      ],
      "description": "Safe error details when ok is false; credential values are never returned."
    }
  },
  "description": "Normalized GHL MCP response envelope.",
  "additionalProperties": false
}

Code Examples

cURL

curl -X POST 'https://portal.madpanda3d.com/api/mcp' \
  -H 'Authorization: Bearer mad_live_***' \
  -H 'Content-Type: application/json' \
  -d '{
  "jsonrpc": "2.0",
  "id": "ghl-conversation-ai-create-agent-example",
  "method": "tools/call",
  "params": {
    "name": "ghl.conversation_ai_create_agent",
    "arguments": {
      "body": {},
      "query": {},
      "tenantId": "tenantId_123",
      "locationId": "locationId_123",
      "strictTenant": false,
      "includeFullConfig": false
    }
  }
}'

Node.js

const response = await fetch('https://portal.madpanda3d.com/api/mcp', {
  method: 'POST',
  headers: {
    Authorization: 'Bearer mad_live_***',
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
  "jsonrpc": "2.0",
  "id": "ghl-conversation-ai-create-agent-example",
  "method": "tools/call",
  "params": {
    "name": "ghl.conversation_ai_create_agent",
    "arguments": {
      "body": {},
      "query": {},
      "tenantId": "tenantId_123",
      "locationId": "locationId_123",
      "strictTenant": false,
      "includeFullConfig": false
    }
  }
}),
});
const data = await response.json();
console.log(data);

Python

import requests

url = 'https://portal.madpanda3d.com/api/mcp'
headers = {
    'Authorization': 'Bearer mad_live_***',
    'Content-Type': 'application/json',
}
payload = {
  "jsonrpc": "2.0",
  "id": "ghl-conversation-ai-create-agent-example",
  "method": "tools/call",
  "params": {
    "name": "ghl.conversation_ai_create_agent",
    "arguments": {
      "body": {},
      "query": {},
      "tenantId": "tenantId_123",
      "locationId": "locationId_123",
      "strictTenant": false,
      "includeFullConfig": false
    }
  }
}
response = requests.post(url, headers=headers, json=payload, timeout=60)
print(response.json())

Example Responses

200 Success

{
  "ok": true,
  "message": "Tool ghl.conversation_ai_create_agent executed successfully.",
  "note": "Response shape varies by MCP tool. Inspect live responses for exact fields."
}

502 Upstream Error

{
  "ok": false,
  "classification": "upstream_http_error",
  "message": "Upstream MCP returned a non-success status."
}