> ## Documentation Index > Fetch the complete documentation index at: https://flowdrop.mintlify.site/llms.txt > Use this file to discover all available pages before exploring further. # Send a chat message > Send a natural language message to the LLM chat backend for a specific workflow. The request includes the current workflow state and optional conversation history so the LLM can generate contextually relevant responses. The LLM response may contain plain text explanations and/or DSL commands in ```flowdrop fenced code blocks. The frontend extracts and previews these commands before execution. ## OpenAPI ````yaml /api-reference/openapi.yaml post /workflows/{id}/chat/messages openapi: 3.0.3 info: title: FlowDrop™ API description: > FlowDrop is a visual workflow editor for AI applications and data processing pipelines. This API provides comprehensive endpoints for managing workflows, node types, pipeline execution, and port configuration. ## Features - **Workflow Management**: Complete CRUD operations for workflows with nodes and edges - **Node Type Discovery**: Browse and search available node processors with metadata - **Pipeline Execution**: Execute workflows with real-time status tracking and job management - **Port Configuration**: Dynamic port compatibility system with data type management - **Real-time Updates**: Track node execution status and pipeline progress - **Import/Export**: Import and export workflows in JSON format - **Validation**: Workflow validation before execution - **Agent Spec Integration**: Import/export Oracle Open Agent Spec flows and execute on compatible runtimes (WayFlow/PyAgentSpec) ## Architecture - **Frontend**: Svelte 5 + XYFlow for visual workflow editing - **Backend**: Drupal 10/11 with custom node processor plugins - **API**: RESTful JSON API with consistent response format - **Storage**: Drupal entity system with workflow versioning ## Authentication API endpoints require Drupal authentication. Use Bearer token or session-based authentication. ## Rate Limiting - Node Discovery: 100 requests/minute - Workflow Operations: 50 requests/minute - Pipeline Execution: 20 requests/minute ## Error Handling All errors return a consistent format with success flag, error message, and optional details. version: 1.0.0 contact: name: FlowDrop Support email: shibinkidd@gmail.com url: https://www.drupal.org/project/issues/flowdrop?categories=All servers: - url: http://localhost:5173/api/flowdrop description: Local development server (Svelte) - url: https://flowdrop.ddev.site/api/flowdrop description: Local Drupal server security: - BearerAuth: [] - SessionAuth: [] tags: - name: System description: System health and status endpoints - name: Node Types description: Node type discovery and metadata endpoints - name: Configuration description: System configuration endpoints including port configuration - name: Workflows description: Workflow CRUD operations - name: Pipeline description: Pipeline execution and monitoring - name: Playground description: Interactive workflow testing and chat interface - name: Interrupts description: | Human-in-the-Loop (HITL) interrupt endpoints for workflow interactions. Interrupts allow workflows to pause execution and request user input. - name: Validation description: Workflow validation - name: Import/Export description: Workflow import and export operations - name: Chat description: | LLM Chat Interface for natural language workflow building. Translates user intent into DSL commands via a backend LLM integration. - name: Agent Spec description: > Oracle Open Agent Spec integration endpoints. Provides bidirectional conversion between FlowDrop workflows and Agent Spec format, plus runtime execution on compatible runtimes (WayFlow, PyAgentSpec). @see https://github.com/oracle/agent-spec paths: /workflows/{id}/chat/messages: post: tags: - Chat summary: Send a chat message description: > Send a natural language message to the LLM chat backend for a specific workflow. The request includes the current workflow state and optional conversation history so the LLM can generate contextually relevant responses. The LLM response may contain plain text explanations and/or DSL commands in ```flowdrop fenced code blocks. The frontend extracts and previews these commands before execution. operationId: sendChatMessage parameters: - name: id in: path description: Workflow ID required: true schema: type: string requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/ChatRequest' examples: simpleMessage: summary: Simple chat message value: message: Add a processing node called "transform" workflowState: nodes: [] edges: [] withHistory: summary: Message with conversation history value: message: Now connect it to the start node workflowState: nodes: - id: node-1 type: start - id: node-2 type: processing edges: [] history: - role: user content: Add a processing node called "transform" - role: assistant content: Done! I've added a processing node named "transform". responses: '200': description: Chat response with LLM-generated content content: application/json: schema: $ref: '#/components/schemas/ChatMessagesResponse' '400': $ref: '#/components/responses/BadRequest' '404': $ref: '#/components/responses/NotFound' '500': $ref: '#/components/responses/InternalServerError' components: schemas: ChatRequest: type: object description: Request payload for sending a chat message required: - message - workflowState properties: message: type: string description: The user's natural language message example: Add a node called "transform" and connect it to the start node workflowState: type: object description: | Serialized current workflow state including nodes and edges. This gives the LLM context about the current workflow being edited. additionalProperties: true history: type: array description: Optional conversation history for multi-turn context items: $ref: '#/components/schemas/ChatHistoryMessage' ChatMessagesResponse: type: object description: Wrapped response for chat message send properties: success: type: boolean example: true data: $ref: '#/components/schemas/ChatResponse' ChatHistoryMessage: type: object description: A single message in the chat conversation history required: - role - content properties: role: $ref: '#/components/schemas/ChatMessageRole' content: type: string description: The message text content example: Add a new processing node called "transform" ChatResponse: type: object description: | Response from the chat endpoint containing the LLM's reply. The content may include markdown formatting and fenced code blocks with DSL commands (```flowdrop blocks). required: - content properties: content: type: string description: | The LLM's response content. May contain markdown and ```flowdrop fenced code blocks with DSL commands. example: | I'll add a processing node and connect it to your start node. ```flowdrop add transform processing connect start:output transform:input ``` conversationId: type: string description: Optional conversation ID for backend session tracking example: conv-abc123 ErrorResponse: allOf: - $ref: '#/components/schemas/ApiResponse' - type: object properties: success: type: boolean example: false error: type: string description: Error message code: type: string description: Error code details: type: object description: Additional error details ChatMessageRole: type: string description: Role of the message sender enum: - user - assistant example: user ApiResponse: type: object properties: success: type: boolean description: Whether the request was successful data: description: Response data (type varies by endpoint) message: type: string description: Response message error: type: string description: Error message (if any) responses: BadRequest: description: Bad request - invalid parameters or request body content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: success: false error: Invalid request parameters code: VALIDATION_ERROR details: field: name message: Name is required NotFound: description: Resource not found content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: success: false error: Resource not found code: NOT_FOUND InternalServerError: description: Internal server error content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: success: false error: Internal server error code: INTERNAL_ERROR securitySchemes: BearerAuth: type: http scheme: bearer bearerFormat: JWT description: JWT token for authentication SessionAuth: type: apiKey in: cookie name: SESS description: Drupal session cookie ````