Discovering Tools

Learn what tools each connector offers and what capabilities are available for a user's connections.

List Tools for a Provider

curl https://api.bluenexus.ai/api/v1/providers/google/tools \
  -H "Authorization: Bearer ACCESS_TOKEN"

Optionally filter by connection:

curl "https://api.bluenexus.ai/api/v1/providers/google/tools?connectionId=conn-abc123" \
  -H "Authorization: Bearer ACCESS_TOKEN"

Response:

[
  {
    "name": "keyword_search",
    "description": "Search across Google Workspace for relevant content",
    "parameters": {
      "type": "object",
      "properties": {
        "query": { "type": "string", "description": "Search query" }
      },
      "required": ["query"]
    },
    "mode": "read"
  },
  {
    "name": "get_calendar_events",
    "description": "Get calendar events for a date range",
    "parameters": {
      "type": "object",
      "properties": {
        "start_date": { "type": "string", "description": "Start date (ISO 8601)" },
        "end_date": { "type": "string", "description": "End date (ISO 8601)" }
      },
      "required": ["start_date"]
    },
    "mode": "read"
  }
]

Tool Properties

Field Description
name Tool identifier
description What the tool does
parameters JSON Schema for the tool's input
mode "read" for read-only, "write" for mutations

Every provider includes a keyword_search tool as a standardized search interface.

Runtime Discovery via MCP

When using the MCP endpoint, call list-connections to discover what's available at runtime:

{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "list-connections",
    "arguments": {}
  },
  "id": "1"
}

This returns the user's active connections with service names and connected accounts. The use-agent tool then dynamically discovers and uses the right tools based on the user's prompt.

Note on Tool Execution

You don't call individual provider tools directly through the MCP endpoint. Instead:

  1. You call use-agent with a natural language prompt
  2. The BlueNexus agent internally discovers and executes the appropriate provider tools
  3. The agent returns the processed result

This is by design — it keeps your context window small and handles the complexity of multi-step tool orchestration.

Next Steps