Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.auvy.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
MCP Examples

​
1) Discover Session Context

Use this first to confirm workspace and brain selection. 
{
  "tool": "get_session",
  "arguments": {}
}

​
2) Browse Then Read

Keyword browse: 
{
  "tool": "search",
  "arguments": {
    "query": "neurons/support"
  }
}
Then read a returned path: 
{
  "tool": "read",
  "arguments": {
    "path": "neurons/abc-123"
  }
}

{
  "tool": "search",
  "arguments": {
    "query": "pathways/customer escalation"
  }
}
Use returned path values with read for full records.

​
4) Delegate To A Neuron

{
  "tool": "delegate",
  "arguments": {
    "path": "neurons/support-agent",
    "message": "Summarize open incidents"
  }
}

​
5) Create Resource

{
  "tool": "create_resource",
  "arguments": {
    "path": "neurons/customer-support-agent",
    "config": {
      "description": "A support neuron that handles billing and refund questions.",
      "neuron_config": {
        "system_prompt": "You are a concise billing support agent.",
        "tools": []
      }
    }
  }
}

​
6) Patch Resource

{
  "tool": "apply_patch",
  "arguments": {
    "kind": "neuron",
    "resource_id": "00000000-0000-4000-8000-000000000000",
    "patch": {
      "patch_mode": "neuron_partial",
      "description": "Handles support issues with concise responses."
    }
  }
}

​
7) Work With History

Search execution history: 
{
  "tool": "grep_history",
  "arguments": {
    "query": "incident"
  }
}
Read history entry: 
{
  "tool": "read_history",
  "arguments": {
    "path": "engrams/engram-uuid"
  }
}

​
8) MCP App Rendering

For receptor paths, read can return the receptor viewer MCP app resource:
  • ui://receptor-viewer/mcp-app.html
Clients that support MCP Apps (for example Claude and ChatGPT) can render that UI.

​
9) Run A Connect Integration Tool

Use direct execution with an integration that is already configured in AUVY, then call connect_execute with integration_id, tool_name, and tool_config. 
{
  "tool": "connect_execute",
  "arguments": {
    "integration_id": "integration-uuid",
    "tool_name": "GMAIL_SEND_EMAIL",
    "tool_config": {
      "to": "alex@example.com",
      "subject": "Status update",
      "body": "The run completed successfully."
    }
  }
}

​
10) Add Brain Context

For brain-scoped search or history tools, include X-Brain-Id in your MCP server config: 
{
  "mcpServers": {
    "auvy": {
      "url": "https://mcp.auvy.ai/mcp",
      "headers": {
        "Authorization": "Bearer ak_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
        "X-Brain-Id": "00000000-0000-0000-0000-000000000000"
      }
    }
  }
}