MCP Trigger

Expose any Tensorify workflow as an MCP tool that external AI agents can discover and call. Works with Claude Desktop, Cursor, Windsurf, and any MCP-compatible client.

The MCP Trigger turns your workflow into a tool in the Model Context Protocol ecosystem. When an external AI agent calls tools/list, it sees your workflow listed with its name, description, and parameter schema. When it calls tools/call, your workflow executes and returns the result.

| Setting | Type | Description | |---------|------|-------------| | Tool Name | string | The identifier agents see (e.g. create_invoice, search_docs) | | Tool Description | text | Helps agents understand when to use this tool | | Input Parameters | JSON Schema | Defines what arguments the tool accepts | | Response Mode | select | use-workflow-response (wait for Return node) or ack-immediately | | Timeout | number | Max seconds to wait for workflow response (default: 30) | | Mock Payload | JSON | Test payload for local development |

The trigger emits an mcp object with:

• mcp.tool_name — The tool name from the MCP call
• mcp.arguments — The full arguments object passed by the agent
• mcp.call_id — Unique identifier for this invocation

Access individual arguments with mcp.arguments.query, mcp.arguments.id, etc.

1. You design a workflow with an MCP Trigger node
2. Configure the tool name, description, and parameter schema
3. Deploy the workflow
4. The tool appears at https://mcp.tensorify.io/v1/triggers
5. External agents authenticate with your API key and call the tool
6. The workflow executes and returns its result to the agent

After deploying, add this to your MCP client config:

{
  "mcpServers": {
    "tensorify-tools": {
      "url": "https://mcp.tensorify.io/v1/triggers",
      "headers": {
        "Authorization": "Bearer tfk_your_api_key_here"
      }
    }
  }
}

Or run tensorify mcp config to generate the config automatically.

A workflow that searches your documentation:

• Tool Name: search_docs
• Description: "Search the documentation knowledge base and return relevant results"
• Parameters:

{
  "type": "object",
  "properties": {
    "query": {
      "type": "string",
      "description": "The search query"
    },
    "limit": {
      "type": "number",
      "description": "Max results to return"
    }
  },
  "required": ["query"]
}

When Claude Desktop calls this tool, it passes the query and receives search results back.

| | MCP Trigger | MCP Tool Provider | |---|---|---| | Direction | Exposes workflow as a tool | Connects workflow to external tools | | Use case | Other agents call your workflow | Your AI agent calls external MCP servers | | Protocol role | Server (provides tools) | Client (consumes tools) |