Vertex

A powerful local AI workflow system with multi-model support and visual workflow orchestration.

Features

Category	Feature	Description
AI Models	Multi-Model Support	Ollama local models and external APIs (DeepSeek, OpenRouter, Tongyi)
Advanced AI	🎨 Multimodal Support	Image analysis and text+image conversations with Gemini 2.5 Pro
	🤔 Reasoning Display	AI thinking process visualization (supports DeepSeek R1)
	🔬 Deep Research	Six-stage research workflow with intelligent analysis
Tools & Search	🔍 Smart Web Search	Multi-engine support (SerpAPI, DuckDuckGo, Bocha AI, etc.)
	Function Tools	Command line execution, web search, financial data tools
Interface	⚡ Streaming Output	Real-time AI response display for better interaction
	Unified CLI	Simple command interface with multiple operation modes
	Desktop Application	Native desktop app with PyWebView integration
Workflow	VertexFlow Engine	Visual workflow orchestration with drag-and-drop nodes
	RAG System	Local Retrieval-Augmented Generation with document processing
Configuration	Smart Configuration	Simplified configuration system with automatic setup
	Document Processing	Support for TXT, MD, PDF, DOCX formats

Quick Start

Requirements

• Python 3.9+
• Ollama (for local models) - Download here

Installation

# Install via pip (recommended)
pip install vertex

# Or install from source
git clone https://github.com/ashione/vertex.git
cd vertex
pip install -e .

Configuration

# Quick setup - Initialize configuration
vertex config init

# Interactive configuration wizard
vertex config

# Check configuration status
vertex config check

Launch

# Standard chat mode (default)
vertex

# Advanced workflow chat with function tools and reasoning
python -m vertex_flow.src.workflow_app --port 7864

# Deep Research analysis tool
vertex deepresearch
# or short form
vertex dr

# VertexFlow workflow mode
vertex workflow

# RAG document Q&A mode
vertex rag --interactive

# Desktop mode
vertex --desktop

Access the Web interface at http://localhost:7860 (or http://localhost:7864 for workflow app)

Usage Guide

CLI Commands

# Standard mode
vertex                    # Launch chat interface
vertex run --port 8080   # Custom port

# Advanced workflow chat mode
python -m vertex_flow.src.workflow_app --port 7864  # With function tools, web search, reasoning

# Deep Research mode
vertex deepresearch       # Start deep research analysis tool
vertex dr --topic "AI trends"  # Direct research from command line
vertex dr --port 8080     # Custom port for web interface

# Workflow mode
vertex workflow           # Visual workflow editor
vertex workflow --port 8080

# Configuration management
vertex config             # Interactive setup
vertex config init        # Quick initialization
vertex config check       # Check status
vertex config reset       # Reset to template

# RAG system
vertex rag --interactive  # Interactive Q&A
vertex rag --query "question"  # Direct query
vertex rag --directory /path/to/docs  # Index documents

# Desktop mode
vertex --desktop          # Desktop application
vertex workflow --desktop # Desktop workflow editor

Deep Research System

The Deep Research tool provides comprehensive analysis through a six-stage workflow:

1. Topic Analysis 🔍 - Initial topic understanding and scope definition
2. Research Planning 📋 - Strategic research approach and methodology
3. Information Collection 📚 - Comprehensive data gathering and source compilation
4. Deep Analysis 🔬 - In-depth examination and critical evaluation
5. Cross Validation ✅ - Verification and fact-checking across sources
6. Summary Report 📄 - Professional research report generation

# Deep Research via API
from vertex_flow.src.deep_research_app import DeepResearchApp

app = DeepResearchApp()
# Configure research parameters and execute

RAG System

from vertex_flow.workflow.unified_rag_workflow import UnifiedRAGSystem

# Create RAG system
rag_system = UnifiedRAGSystem()

# Index documents
documents = ["document1.txt", "document2.pdf"]
rag_system.index_documents(documents)

# Query the knowledge base
answer = rag_system.query("What is the main topic?")
print(answer)

Function Tools

# Access various function tools through service
from vertex_flow.workflow.service import VertexFlowService

service = VertexFlowService()
cmd_tool = service.get_command_line_tool()      # Command execution
web_tool = service.get_web_search_tool()        # Smart web search (SerpAPI/DuckDuckGo/Bocha etc.)
finance_tool = service.get_finance_tool()       # Financial data retrieval

# Tools integrate seamlessly with AI workflows, supporting streaming and reasoning

Basic Workflow

from vertex_flow.workflow.vertex.vertex import SourceVertex
from vertex_flow.workflow.workflow import Workflow
from vertex_flow.workflow.context import WorkflowContext

def source_func(inputs, context):
    return {"text": "Hello, Vertex Flow!"}

context = WorkflowContext()
workflow = Workflow(context)
source = SourceVertex(id="source", task=source_func)
workflow.add_vertex(source)
workflow.execute_workflow()

Configuration

Quick Setup

After installing the vertex package, use these commands for quick setup:

# Initialize configuration file
vertex config init

# Interactive configuration wizard
vertex config

# Check configuration status
vertex config check

# Reset configuration
vertex config reset

Manual Configuration

Configuration file is located at ~/.vertex/config/llm.yml. You can edit this file directly.

Environment Variables

Set API keys for external models:

export llm_deepseek_sk="your-deepseek-key"
export llm_openrouter_sk="your-openrouter-key"
export llm_tongyi_sk="your-tongyi-key"
export web_search_serpapi_api_key="your-serpapi-key"
export web_search_bocha_sk="your-bocha-key"

Configuration Priority

1. User configuration file: ~/.vertex/config/llm.yml
2. Environment variables
3. Package default configuration

Documentation

📖 User Guides

• Complete CLI Usage Guide - Full CLI command reference and MCP integration
• Desktop Application Guide - Desktop app usage
• Workflow Chat App Guide - Advanced chat with function tools and reasoning
• 🎨 Multimodal Features Guide - Image analysis and text+image conversations
• 🔍 Web Search Configuration - Multi-engine search setup
• MCP Integration Guide - Model Context Protocol support
• Troubleshooting Guide - Common issues and solutions

🔧 Technical Documentation

• Function Tools Guide - Complete function tools reference
• Workflow Chain Calling - Chain workflow execution
• RAG System Overview - Retrieval-Augmented Generation
• Document Update Mechanism - Incremental updates and deduplication
• Deduplication Features - Smart document deduplication
• Configuration Unification - Unified configuration system

🎯 Development & Maintenance

• Publishing Guide - Package publishing and version management
• Pre-commit Checks - Code quality and automated checks

Examples

# Function tools examples
cd vertex_flow/examples
python command_line_example.py   # Command line tool
python web_search_example.py     # Web search tool  
python finance_example.py        # Finance tool
python rag_example.py            # RAG system
python deduplication_demo.py     # Deduplication

Development

# Run pre-commit checks
./scripts/precommit.sh

# Version management
python scripts/version_bump.py

License

See LICENSE file for details.