> ## Documentation Index
> Fetch the complete documentation index at: https://www.voiceflow.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Execute a MCP tool with MCP

> Use voiceflow_mcp_tool with get_schema to discover inputs, then execute to run a saved MCP tool without a conversation.

## When to use this tool

Use `voiceflow_mcp_tool` with `get_schema` to discover inputs, then `execute` to run a saved MCP tool without a conversation. This page covers those two operations; other management operations on the tool remain available.

## Access and effects

Use an authenticated Voiceflow MCP connection with access to the project and environment. Schema discovery requires the corresponding read permission; execution requires execute permission. Calls use saved draft resources and can change external systems. Obtain authorization for the intended external action before executing it.

## Inputs

The current MCP surface uses an `operation` object with `name` and `payload`.

| Field | Required | Meaning |
| --- | --- | --- |
| `operation.name` | Yes | `get_schema` or `execute` |
| `operation.payload.projectID` | Yes | Project to access |
| `operation.payload.environmentAlias` | Yes | Selected environment ID or alias |
| `operation.payload.toolID` | Yes | Saved resource ID |
| `operation.payload.body.inputs` | For execution inputs | Values from the discovered schema; an empty map is appropriate only when the tool needs no inputs |

For schema discovery, omit `body`. Functions and API tools require all declared inputs as strings; other tools follow their input schema.

## Result

Schema discovery describes the accepted inputs. Function schemas also describe output variables and exit paths; integration schemas list catalog actions and connection status; MCP tool schemas include the server ID and JSON Schema. Execution returns the [shared result fields](/docs/developers/tools/execution-behavior) and resource-specific output inside the MCP response. Inspect tool errors as well as the execution `success` field.

## Invocation example

Tool: `voiceflow_mcp_tool`. This is the argument object for execution; replace identifiers and fill `inputs` from `get_schema` before calling.

```json
{
  "operation": {
    "name": "execute",
    "payload": {
      "projectID": "PROJECT_ID",
      "environmentAlias": "ENVIRONMENT",
      "toolID": "RESOURCE_ID",
      "body": {
        "inputs": {}
      }
    }
  }
}
```

## Related tools

[Run tools without a conversation](/docs/developers/tools/execute) · [Execution behavior](/docs/developers/tools/execution-behavior) · [MCP overview](/docs/mcp/overview) · [REST execution reference](/docs/api-reference/mcp-tool/execute-mcp-tool) · [REST schema reference](/docs/api-reference/mcp-tool/get-mcp-tool-schema)

## Failures and follow-up

For unknown or missing inputs, refresh the schema. For missing credentials, repair the saved connection instead of retrying. A tool error is different from an MCP transport failure. After a timeout, verify whether the external action occurred before repeating it. See the [execution guide’s troubleshooting](/docs/developers/tools/execute#troubleshooting).