Implementation:Langgenius Dify Debug Models

Knowledge Sources	Langgenius_Dify
Domains	Frontend, Type Definitions, LLM Configuration, Debugging
Last Updated	2026-02-08 00:00 GMT

Overview

Defines TypeScript types and enums for LLM model configuration, prompt management, completion parameters, dataset retrieval settings, debug requests/responses, and feedback handling in the Dify frontend.

Description

web/models/debug.ts is a central type definition module for the Dify application configuration and debugging system. It provides the type infrastructure for the "Debug and Preview" feature of the application builder, covering the full spectrum of LLM configuration options.

Key type groups include:

Prompt system: PromptMode enum (simple/advanced), PromptRole enum (system/user/assistant), PromptItem, PromptVariable, ChatPromptConfig, CompletionPromptConfig
Model configuration: ModelConfig (comprehensive frontend model configuration), CompletionParams, ModelId
Feature configurations: MoreLikeThisConfig, SuggestedQuestionsAfterAnswerConfig, SpeechToTextConfig, TextToSpeechConfig, CitationConfig, ModerationConfig, AnnotationReplyConfig, RetrieverResourceConfig, AgentConfig
Dataset configurations: DatasetConfigs with retrieval model, reranking settings, score thresholds, metadata filtering
Debug/log types: DebugRequestBody, DebugResponse, DebugResponseStream, FeedBackRequestBody, FeedBackResponse
Session types: LogSessionListQuery, LogSessionListResponse, LogSessionDetailResponse

Usage

Import types from this module when building configuration UIs, debug panels, or handling model-related API responses:

import type { ModelConfig, PromptVariable, CompletionParams } from '@/models/debug'
import { PromptMode, PromptRole } from '@/models/debug'

Code Reference

Source Location

Repository: Langgenius_Dify
File: web/models/debug.ts
Lines: 1-272

Signature

export type Inputs = Record<string, string | number | object | boolean>

export enum PromptMode { simple = 'simple', advanced = 'advanced' }
export enum PromptRole { system = 'system', user = 'user', assistant = 'assistant' }

export type PromptVariable = {
  key: string; name: string; type: string;
  default?: string | number; required?: boolean;
  options?: string[]; max_length?: number;
}

export type CompletionParams = {
  max_tokens: number; temperature: number; top_p: number;
  presence_penalty: number; frequency_penalty: number; stop?: string[];
}

export type ModelConfig = {
  provider: string; model_id: string; mode: ModelModeType;
  configs: PromptConfig; opening_statement: string | null;
  more_like_this: MoreLikeThisConfig | null;
  dataSets: any[]; agentConfig: AgentConfig;
  // ... many more optional feature configs
}

export type DatasetConfigs = {
  retrieval_model: RETRIEVE_TYPE;
  reranking_model: { reranking_provider_name: string; reranking_model_name: string };
  top_k: number; score_threshold: number | null | undefined;
  datasets: { datasets: { enabled: boolean; id: string }[] };
}

export type DebugRequestBody = { inputs: Inputs; query: string; completion_params: CompletionParams; model_config: ModelConfig }
export type DebugResponse = { id: string; answer: string; created_at: string }

export type ModerationConfig = { enabled: boolean; type?: string; config?: { keywords?: string; ... } }

Import

import type { ModelConfig, PromptVariable, DatasetConfigs, ModerationConfig } from '@/models/debug'
import { PromptMode, PromptRole } from '@/models/debug'

I/O Contract

Inputs

Name	Type	Required	Description
(none)	-	-	This module only exports type definitions and enums; it has no runtime inputs

Outputs

Name	Type	Description
PromptMode	enum	Enum with values `simple` and `advanced` for prompt editing modes
PromptRole	enum	Enum with values `system`, `user`, `assistant` for chat prompt roles
ModelConfig	type	Comprehensive frontend model configuration including provider, prompt, features, and dataset settings
CompletionParams	type	LLM generation parameters: temperature, top_p, max_tokens, stop sequences, penalties
DatasetConfigs	type	RAG retrieval configuration with reranking, scoring, and metadata filtering options
DebugRequestBody	type	Request body for debug/preview API calls
ModerationConfig	type	Content moderation configuration with keyword and API-based options
PromptVariable	type	Variable definition for prompt templates with key, type, options, and validation

Usage Examples

import type { ModelConfig, CompletionParams, PromptVariable } from '@/models/debug'
import { PromptRole, PromptMode } from '@/models/debug'

// Build a model configuration for the debug panel
const modelConfig: ModelConfig = {
  provider: 'openai',
  model_id: 'gpt-3.5-turbo',
  mode: 'chat',
  prompt_type: PromptMode.advanced,
  configs: {
    prompt_template: 'You are a helpful assistant.',
    prompt_variables: [],
  },
  chat_prompt_config: {
    prompt: [{ role: PromptRole.system, text: 'You are a helpful assistant.' }],
  },
  opening_statement: 'Hello! How can I help you today?',
  more_like_this: null,
  // ... other required fields
}

// Define prompt variables
const variables: PromptVariable[] = [
  { key: 'name', name: 'User Name', type: 'string', required: true },
  { key: 'language', name: 'Language', type: 'select', options: ['English', 'Chinese'] },
]

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment