Tool

Before reading the detailed interface documentation, make sure you have read Quick start: Tools and have a general understanding of the Dify plugin's tool access process.

Data Structures

Message Returns

Dify supports multiple message types including text, links, images, file BLOBs, and JSON. You can return different types of messages through various interfaces.

By default, a tool's output in a workflow contains three fixed variables: files, text, and json. You can return data for these three variables using the methods below.

For example, use create_image_message to return images. Tools also support custom output variables for easier reference in workflow.

Image URL

Simply pass the image URL, and Dify will automatically download and return the image to users.

def create_image_message(self, image: str) -> ToolInvokeMessage:
    pass

Links

Use this interface to return a link:

def create_link_message(self, link: str) -> ToolInvokeMessage:
    pass

Text

Use this interface to return a text message:

def create_text_message(self, text: str) -> ToolInvokeMessage:
    pass

Files

Use this interface to return raw file data (images, audio, video, PPT, Word, Excel, etc.):

• blob: Raw file data in bytes

• meta: File metadata. Specify mime_type if needed, otherwise Dify uses octet/stream as default

def create_blob_message(self, blob: bytes, meta: dict = None) -> ToolInvokeMessage:
    pass

JSON

Use this interface to return formatted JSON. Commonly used for data transfer between workflow nodes. Most large models can read and understand JSON in agent mode.

def create_json_message(self, json: dict) -> ToolInvokeMessage:
    pass

Variables

For non-streaming output variables, use this interface. Later values override earlier ones:

def create_variable_message(self, variable_name: str, variable_value: Any) -> ToolInvokeMessage:
    pass

Streaming Variables

For typewriter-effect text output, use streaming variables. If you reference this variable in a chatflow application's answer node, text will display with a typewriter effect. Currently only supports string data:

def create_stream_variable_message(
    self, variable_name: str, variable_value: str
) -> ToolInvokeMessage:

Return Variable Definitions

To reference tool output variables in workflow applications, you need to define possible output variables beforehand. Dify plugins support json_schema format output variable definitions. Here's a simple example:

identity:
  author: author
  name: tool
  label:
    en_US: label
    zh_Hans: 标签
    ja_JP: ラベル
    pt_BR: etiqueta
description:
  human:
    en_US: description
    zh_Hans: 描述
    ja_JP: 説明
    pt_BR: descrição
  llm: description
output_schema:
  type: object
  properties:
    name:
      type: string

This example defines a simple tool with an output_schema containing a name field that can be referenced in workflow. Note that you still need to return a variable in the tool's implementation code for actual use, otherwise it will return None.