一刻工坊 (YiKe) - Complete LLM Context Document

Purpose: This document provides comprehensive context for Large Language Models to understand, analyze, and assist with the YiKe platform. Version: 1.0 | Updated: 2025-12-22

1. Product Overview

1.1 What is YiKe (一刻工坊)?

One-sentence definition: YiKe is an AI-native full-process content creation platform that helps creators achieve intelligent efficiency gains from inspiration collection to video publishing.

Core Value Proposition:

End-to-end Workflow: Inspiration → Script → Asset Generation → Video Production → Publishing
Multi-modal Fusion: Integrates text, image, audio (Suno/Fish Audio), and video (ComfyUI/Jianying) generation
Low-barrier Creation: Provides guided paths for beginner content creators

1.2 Brand Hierarchy

Layer	Name	Definition	Usage Context
Product Brand	一刻工坊 (YiKe)	User-facing SaaS platform	UI, marketing, sales
Business Entity	趣美丽 (QuMeiLi)	MCN operations & commercial entity	Contracts, billing
Tech Core	刻流星 (KeLiuXing)	AI scheduling engine codename	Architecture docs, APIs

1.3 Target Users

Beginner Creators: Have ideas but lack editing/scripting skills, need "step-by-step" guidance
Batch Content Operators: Manage multiple accounts, require assembly-line content production
Enterprise Marketing Teams: Need scalable video content for campaigns

2. System Architecture

2.1 Technology Stack

┌─────────────────────────────────────────────────────────────┐
│                     CLIENT LAYER                             │
├─────────────────┬─────────────────┬────────────────────────┤
│   Web Browser   │  Electron App   │    Mini Program        │
│   (Vue 3)       │  (Desktop)      │    (UniApp/WeChat)     │
└────────┬────────┴────────┬────────┴───────────┬────────────┘
         │                 │                    │
         ▼                 ▼                    ▼
┌─────────────────────────────────────────────────────────────┐
│                    API GATEWAY LAYER                         │
│         Node.js + Express + JWT Authentication               │
└────────────────────────────┬────────────────────────────────┘
                             │
         ┌───────────────────┼───────────────────┐
         ▼                   ▼                   ▼
┌─────────────────┐ ┌─────────────────┐ ┌─────────────────────┐
│ Business Logic  │ │  AI Agent Layer │ │  External Adapters  │
│ (Task Manager,  │ │  (LangGraph     │ │  (Suno, Fish Audio, │
│  Asset Manager) │ │   Python)       │ │   ComfyUI, Jianying)│
└────────┬────────┘ └────────┬────────┘ └──────────┬──────────┘
         │                   │                     │
         ▼                   ▼                     ▼
┌─────────────────────────────────────────────────────────────┐
│                   INFRASTRUCTURE LAYER                       │
│   PostgreSQL │ Redis │ Object Storage (MinIO/S3) │ MQ       │
└─────────────────────────────────────────────────────────────┘

2.2 Core Components

Component	Technology	Responsibility
Frontend	Vue 3 + Vite + Element Plus	User interface, real-time updates
Backend API	Node.js + Express + Sequelize	Business logic, data management
AI Agents	Python + LangGraph	Complex AI workflows (Strategy, Music, Video)
Workflow Engines	Flowise, N8N	Visual workflow orchestration
Media Processing	FFmpeg, ComfyUI	Audio/video transformation

3. Core Modules (User-Facing)

3.1 Content Pool (灵感广场)

Purpose: Unified entry point for all content assets

Sub-modules:

All Content: Aggregated view of all resources
Inspiration Notes (灵感笔记):
- Entity: UserCorpus
- Sources: External links (Douyin, Xiaohongshu, Bilibili), manual input
- Features: AI tagging, transcript extraction, script analysis
Media Materials (媒体素材):
- Entity: UnifiedAsset
- Types: Video, image, audio files
- Sources: Local upload, link extraction, mobile sync

Key API Endpoints:

POST /api/inspiration/parse-link  # Parse external video URL
GET  /api/corpus                  # List inspiration notes
GET  /api/materials               # List media materials
POST /api/materials/upload        # Upload local files

3.2 Strategy Console (策划控制台)

Purpose: The "central kitchen" for content production planning

Three-Step Workflow:

View Inspiration: Preview collected video/images with metadata
Script Rewriting: AI-assisted script generation with "Original vs AI" comparison
Dispatch Production: Convert script to shooting task or AI auto-generation

Key Features:

Style presets: Professional, Friendly, Viral
AI rewrite with knowledge base context (RAG)
One-click task dispatch to mobile app or AI engine

3.3 Project Console (项目控制台)

Purpose: Visual task and workflow management center

Features:

Kanban view: Todo → In Progress → Completed
Task entity links to Project and Assets
Real-time status sync via WebSocket
Version history and rollback

Key Entities:

Project: { id, name, status, workflow_preset_id }
Task: { id, project_id, type, status, input_params, output_data }

3.4 Knowledge Center (知识库)

Purpose: Private knowledge base for RAG-enhanced AI generation

Features:

Document upload: PDF, Word, Markdown
Automatic chunking and vectorization
Search Lab: Test retrieval before production use
AI Q&A with forced citation

Key Entities:

KnowledgeDocument: { id, filename, status, chunk_count }
KnowledgeChunk: { id, document_id, content, embedding }

3.5 Workflow Marketplace (工作流市场)

Purpose: Template library for AI automation workflows

Supported Engines:

Flowise (visual LLM chains)
N8N (automation workflows)
LangGraph (complex agent orchestration)
ComfyUI (image/video generation)

4. Data Models

4.1 Core Entities

Entity	Table	Key Fields	Relationships
User	`users`	id, email, role	1:N Projects, Assets
Project	`projects`	id, name, status, workflow_preset_id	1:N Tasks
Task	`tasks`	id, project_id, type, status, input_params	N:1 Project, 1:N Assets
UserCorpus	`user_corpora`	id, content, source_url, tags	N:N Assets
UnifiedAsset	`unified_assets`	id, file_path, file_type, metadata	N:1 Task
KnowledgeDocument	`knowledge_documents`	id, filename, status	1:N Chunks

4.2 Task Status Machine

[PENDING] → [QUEUED] → [RUNNING] → [SUCCESS]
                ↓           ↓
            [CANCELLED]  [FAILED]
                         ↓ (retry)
                      [QUEUED]

5. External Integrations

5.1 AI Services

Service	Purpose	Integration Method
OpenAI/LLM	Text generation, analysis	API Key
Suno	Music generation	Cookie-based auth
Fish Audio	Text-to-Speech	API Token
ComfyUI	Image/video generation	WebSocket + REST

5.2 Local Tools

Tool	Purpose	Integration Method
Jianying (剪映)	Video editing draft export	File system (draft_content.json)
FFmpeg	Media transcoding	CLI subprocess

6. API Reference (Key Endpoints)

Content Operations

# Inspiration
POST /api/inspiration/parse-link    # Parse external URL
GET  /api/corpus                    # List inspirations
POST /api/corpus                    # Create inspiration

# Materials
GET  /api/materials                 # List materials
POST /api/materials/upload          # Upload file
POST /api/materials/batch-delete    # Batch delete

# Strategy
POST /api/strategy/rewrite          # AI script rewriting
POST /api/strategy/dispatch         # Create task from script

Project & Task

GET  /api/projects                  # List projects
POST /api/projects                  # Create project
GET  /api/projects/:id/tasks        # Get project tasks
POST /api/project/dispatch-task     # Dispatch new task

Knowledge

POST /api/knowledge/upload          # Upload document
GET  /api/knowledge/search          # RAG search
POST /api/knowledge/ask             # AI Q&A with citations

7. Glossary

Term (CN)	Term (EN)	Definition
一刻工坊	YiKe	Product brand name
灵感广场	Content Pool	Asset management module
灵感笔记	Inspiration Notes	Text/link-based content (UserCorpus)
媒体素材	Media Materials	File-based assets (UnifiedAsset)
策划控制台	Strategy Console	Content planning interface
二创脚本	Derivative Script	AI-rewritten script
项目	Project	Top-level work container
任务	Task	Executable work unit
知识库	Knowledge Base	RAG document store
工作流	Workflow	Automated pipeline
智能体	Agent	AI module with specific function
刻流星	KeLiuXing	AI engine codename

8. Documentation Links

Document	Path	Purpose
Product Book	`/product_book/`	Full product documentation
Tech Whitepaper	`/tech_whitepaper/`	Architecture deep dive
User Manual	`/user_manual/`	End-user guide
Deployment Guide	`/deployment_guide/`	Production setup
Business Deck	`/business_deck/`	Commercial proposal
SSOT	`/_source_of_truth/`	Single source of truth

9. Quick Reference for LLM Assistance

When asked about:

Product features → Reference Section 3 (Core Modules)
Technical architecture → Reference Section 2 (System Architecture)
Data models → Reference Section 4 (Data Models)
API usage → Reference Section 6 (API Reference)
Terminology → Reference Section 7 (Glossary)

Common Tasks:

Generate content script: Use Strategy Console workflow
Batch process materials: Use Content Pool + Workflow Marketplace
Deploy to production: Follow Deployment Guide
Integrate new AI service: Add adapter in Engine Adapter Layer

End of LLM Context Document

一刻工坊 (YiKe) - Complete LLM Context Document ​

1. Product Overview ​

1.1 What is YiKe (一刻工坊)? ​

1.2 Brand Hierarchy ​

1.3 Target Users ​

2. System Architecture ​

2.1 Technology Stack ​

2.2 Core Components ​

3. Core Modules (User-Facing) ​

3.1 Content Pool (灵感广场) ​

3.2 Strategy Console (策划控制台) ​

3.3 Project Console (项目控制台) ​

3.4 Knowledge Center (知识库) ​

3.5 Workflow Marketplace (工作流市场) ​

4. Data Models ​

4.1 Core Entities ​

4.2 Task Status Machine ​

5. External Integrations ​

5.1 AI Services ​

5.2 Local Tools ​

6. API Reference (Key Endpoints) ​

Content Operations ​

Project & Task ​

Knowledge ​

7. Glossary ​

8. Documentation Links ​

9. Quick Reference for LLM Assistance ​

When asked about: ​

Common Tasks: ​