Weather Syntax (Full Version)

TradingFlow Weather Language defines the complete structure and specification for workflow files (.wt files). A Weather file is essentially a superset of JSON format, containing all information about a workflow: metadata, node configurations, connection relationships, and more.

📌 Version Note: This document describes the Full Version, which includes all fields needed for frontend display. The Essential Version used by Agent generation and backend storage is documented in Weather Syntax Comparison.

.wt File Structure

A complete Weather workflow file consists of three top-level fields:

{
    "thumbnailUrl": "...",
    "name": "...",
    "nodes": [...],
    "edges": [...]
}

Below we will explain each field in detail.

Workflow Metadata

thumbnailUrl

The thumbnail image of the workflow, using Base64-encoded image data. This thumbnail is displayed in workflow lists and preview interfaces.

"thumbnailUrl": "data:image/png;base64,iVBORw0KGgoAAAANS..."

Format Specifications:

  • Data URI format: data:image/png;base64,<encoded-data>

  • Supported image formats: PNG, JPEG, WebP

  • Recommended size: 800x600 pixels

  • Automatically generated when users save workflows

name

The name of the workflow, used to identify and distinguish different workflows.

"name": "New Flow"

Naming Rules:

  • Supports English and international characters

  • Length limit: 1-100 characters

  • Spaces and special characters allowed

  • Default name is "New Flow"

Node Definitions (nodes)

The nodes array contains complete definitions of all nodes in the workflow. Each node is an independent functional unit responsible for specific data processing tasks.

Node Basic Structure

{
    "position": {
        "x": 100,
        "y": 100
    },
    "id": "x_listener_node_1",
    "type": "x_listener_node",
    "className": "",
    "data": {...},
    "width": 343,
    "height": 464,
    "selected": false
}

position

The node's position coordinates on the canvas, used for visual rendering.

"position": {
    "x": 100,  // X-axis coordinate (pixels)
    "y": 100   // Y-axis coordinate (pixels)
}

  • Origin is at the top-left corner of the canvas

  • Updated when users drag nodes

  • Final position is saved

id

The unique identifier of the node, used to reference the node throughout the workflow.

"id": "x_listener_node_1"

Naming Convention:

  • Format: <node_type>_<index>

  • Must be unique within the workflow

  • Automatically generated when creating nodes

  • Used for edge connection references

type

The type identifier of the node, determining its functionality and behavior.

"type": "x_listener_node"

Common Node Types:

  • x_listener_node - X (Twitter) Listener Node

  • ai_model_node - AI Model Node

  • buy_node - Buy Trade Node

  • sell_node - Sell Trade Node

  • code_node - Code Execution Node

  • binance_price_node - Binance Price Node

  • dataset_output_node - Dataset Output Node

width & height

The display dimensions of the node on the canvas (pixels).

"width": 343,
"height": 464

  • Automatically calculated based on node content

  • Different node types have different default sizes

  • Users can manually adjust size

selected

Indicates whether the node is currently selected, used for UI interaction states.

"selected": false

  • true - Node is selected (highlighted)

  • false - Node is not selected (normal display)

Node Data

The data object contains the core configuration information and business logic of the node.

"data": {
    "title": "Trump Twitter Listener",
    "description": "Monitors tweets from @realDonaldTrump",
    "collection": "input",
    "inputs": [...],
    "outputs": [...],
    "id": "x_listener_node_1",
    "edges": [...],
    "menuItems": [...],
    "isDeepEdit": true,
    "isFlowExecuting": false,
    "isStopping": false,
    "signals": []
}

title & description

The display title and description information of the node.

"title": "Trump Twitter Listener",
"description": "Monitors tweets from @realDonaldTrump"

  • title - Name displayed at the top of the node card

  • description - Brief explanation of node functionality

  • Customizable by users

collection

The functional category the node belongs to.

"collection": "input"

Category Types:

  • input - Data input nodes (e.g., listeners, data sources)

  • compute - Computing nodes (e.g., AI models, code execution)

  • trade - Trading operation nodes (e.g., buy, sell)

  • core - Core function nodes (e.g., data output)

inputs

The input parameter configuration array, defining the data and parameters the node receives.

"inputs": [
    {
        "id": "accounts",
        "title": "Accounts",
        "type": "paragraph",
        "inputType": "text",
        "required": true,
        "placeholder": "Enter X accounts to monitor (comma separated)",
        "handle": {
            "color": "sky"
        },
        "value": ["realDonaldTrump"],
        "options": []
    }
]

Input Field Descriptions:

  • id - Identifier of the input parameter

  • title - Display name of the parameter

  • type - Data type (paragraph, text, number, select, searchSelect, button, object, paramMatrix)

  • inputType - Input control type

  • required - Whether the field is required

  • placeholder - Input hint text

  • handle - Connection handle configuration (color, style)

  • value - Current value of the parameter

  • options - List of options (for dropdown selection)

Handle Color Conventions:

  • sky - Data flow type (blue)

  • emerald - Trading type (green)

  • amber - Configuration type (orange)

  • rose - Output type (pink)

outputs

The output definition array, specifying the data outputs produced by the node.

"outputs": [
    {
        "id": "latest_tweets",
        "title": "Latest Tweets",
        "type": "object",
        "handle": {
            "color": "sky"
        },
        "isDeleted": false
    }
]

Output Field Descriptions:

  • id - Identifier of the output

  • title - Display name of the output

  • type - Output data type (text, number, object, array)

  • handle - Connection handle configuration

  • isDeleted - Flag indicating if output is deleted

edges

All connection relationships the node participates in. This is an array containing complete edge information, used for data flow management between nodes.

"edges": [
    {
        "id": "edge-1759136854519-ra1cx77x1",
        "type": "default",
        "source": "x_listener_node_1",
        "target": "ai_model_node_1",
        "sourceHandle": "latest_tweets",
        "targetHandle": "prompt"
    }
]

Note: The edges array here is a node-level reference. Complete edge definitions are in the top-level edges field.

Node State Flags

"isDeepEdit": true,
"isFlowExecuting": false,
"isStopping": false,
"signals": []

  • isDeepEdit - Whether the node is in deep edit mode

  • isFlowExecuting - Whether the workflow is executing

  • isStopping - Whether the workflow is stopping

  • signals - Signal data received by the node at runtime

Edge Definitions (edges)

The edges array defines the connection relationships between nodes, describing the flow paths of data in the workflow.

Edge Basic Structure

{
    "id": "edge_ai_model_node_1_ai_response-handle_to_buy_node_1_buy_token-handle",
    "animated": true,
    "source": "ai_model_node_1",
    "target": "buy_node_1",
    "sourceHandle": "ai_response",
    "targetHandle": "buy_token"
}

id

The unique identifier of the edge, used for referencing and managing connections.

"id": "edge_ai_model_node_1_ai_response-handle_to_buy_node_1_buy_token-handle"

Naming Format:

  • Pattern: edge_<source_node>_<source_handle>_to_<target_node>_<target_handle>

  • Or simplified format: edge-<timestamp>-<random>

  • Must be globally unique

type & animated

Display style configuration of the edge.

"type": "default",
"animated": true

  • type - Type of edge (default, step, smoothstep, straight)

  • animated - Whether to display flow animation (true indicates data is being transmitted)

source & target

The source and target nodes connected by the edge.

"source": "ai_model_node_1",
"target": "buy_node_1"

  • source - ID of the data output node

  • target - ID of the data receiving node

  • Must reference existing node IDs

sourceHandle & targetHandle

The specific input/output ports connected by the edge.

"sourceHandle": "ai_response",
"targetHandle": "buy_token"

  • sourceHandle - Output parameter ID of the source node

  • targetHandle - Input parameter ID of the target node

  • Must match input/output IDs in node definitions

  • Does not include -handle suffix (this is Weather language specification)

Complete Example Walkthrough

Let's understand the practical application of Weather syntax through a complete workflow example:

Workflow Overview

This is an automated trading strategy based on Trump's tweets:

  1. X Listener Node - Monitors tweets from @realDonaldTrump

  2. AI Model Node - Analyzes tweet content to determine buy signals

  3. Buy Node - Executes buy operations based on AI analysis

  4. Binance Price Node - Monitors BTC/USDT price

  5. Code Node - Calculates 5x buy price as sell target

  6. Sell Node - Executes sell when price reaches 5x

  7. Dataset Output Node - Records all trades to Google Sheets

Data Flow

Tweet Monitor → AI Analysis → Buy Trade → Trade Record
                  ↓
Price Monitor → Price Calc → Sell Trade → Trade Record

Key Features

Multiple Input Sources:

  • Dataset Output node receives trade records from both Buy and Sell nodes

  • Data convergence through two independent edges

Conditional Branching:

  • AI analysis result determines whether to trigger buy

  • Price calculation result determines whether to trigger sell

Parameter Passing:

  • AI output directly connects to Buy node's buy_token input

  • Code output connects to Sell node's sell_token and amount_in_human_readable inputs

Syntax Validation Rules

Required Field Validation

All nodes must include the following fields:

  • id, type, position, data

All edges must include the following fields:

  • id, source, target, sourceHandle, targetHandle

Reference Integrity

  • Edge source and target must reference existing node IDs

  • sourceHandle must exist in source node's outputs

  • targetHandle must exist in target node's inputs

Type Compatibility

  • Output type and input type connected by an edge must be compatible

  • For example: object type output can connect to object type input

Uniqueness Constraints

  • Node IDs must be globally unique

  • Edge IDs must be globally unique

Best Practices

Node Naming

Use meaningful titles and descriptions to understand workflow logic:

"title": "Trump Twitter Listener",
"description": "Monitors tweets from @realDonaldTrump"

Parameter Configuration

Set reasonable default values and hints for required parameters:

"required": true,
"placeholder": "Enter X accounts to monitor (comma separated)",
"value": ["realDonaldTrump"]

Edge Organization

Use descriptive edge IDs for easier debugging and maintenance:

"id": "edge_ai_model_node_1_ai_response_to_buy_node_1_buy_token"

State Management

Use animated flag appropriately to indicate data flow status:

"animated": true  // Indicates active data transmission on this edge

Extension and Compatibility

Weather Language is designed as an extensible format:

Backward Compatibility:

  • New fields won't break workflows from older versions

  • Unknown fields are safely ignored

Custom Extensions:

  • Can add custom fields in the data object

  • Node developers can define proprietary configuration parameters

Version Management:

  • Support multiple format versions by adding version field

  • Platform automatically migrates old version workflows

FAQ

Q: Why don't handle names include the -handle suffix?

A: This is by design in the Weather language specification. Earlier versions used the -handle suffix, but it was removed in newer versions to simplify syntax and improve readability. The frontend automatically adds the suffix when needed.

Q: How to handle optional inputs for nodes?

A: Set the input's required field to false and provide a reasonable default value or leave it empty.

Q: What is the purpose of the edge type field?

A: It determines the rendering style of the edge on the canvas. default is a standard curve, straight is a straight line, and step is a stepped line.

Q: What is the purpose of the positionAbsolute field?

A: This is the calculated absolute position, considering canvas zoom and pan. Usually maintained automatically by the frontend.

Next Step: Check out Node Details to learn detailed parameters and usage for each node type.

PreviousNodes and WorkflowsNextWeather Syntax Version Comparison

Last updated