OriCore

Powerful AI Engine Library - Build Smart Assistants in 5 Lines of Code

What is OriCore?

OriCore is a fully-featured AI Engine Library that lets you easily integrate powerful AI capabilities into any application.

Support for 40+ AI providers (OpenAI, Claude, DeepSeek, Zhipu AI, etc.)
Complete built-in tool system (file read/write, code search, shell commands, network requests)
Session management + Context compression
MCP protocol + Skill system for unlimited extensions
5 professional interaction modes (brainstorm, plan, review, debug, default)

Build AI-powered products with just 5 lines of code:

import { createEngine } from 'oricore';

const engine = createEngine({ productName: 'MyApp', version: '1.0.0' });
await engine.initialize({ model: 'deepseek/deepseek-chat', provider: { deepseek: { apiKey: 'your-key' } } });

const result = await engine.sendMessage({ message: 'Analyze this project structure', write: true });
console.log(result.data.text);

Quick Start

npm install oricore ai

import { createEngine } from 'oricore';

const engine = createEngine({ productName: 'MyAIAssistant', version: '1.0.0' });
await engine.initialize({
  model: 'deepseek/deepseek-chat',
  provider: { deepseek: { apiKey: 'your-api-key' } },
});

const result = await engine.sendMessage({
  message: 'Create a TypeScript function to calculate Fibonacci',
  write: true,
});

console.log(result.data.text);
await engine.shutdown();

📖 5-Minute Quick Start Guide

Why OriCore?

What You Need	OriCore	Others
Build an AI assistant fast	✅ 5 lines of code	❌ 100+ lines of setup
Work with any AI provider	✅ 40+ providers unified	⚠️ Provider-specific SDKs
Ready-to-use tools	✅ File, shell, search, web	❌ Build everything yourself
Save on API costs	✅ Auto context compression	❌ Pay for full context every time
Extend with plugins	✅ MCP + Skills	❌ Complex or impossible
Switch AI providers	✅ Change 1 line of code	❌ Rewrite integration code

Built for developers who need to ship AI features fast:

5-minute setup - Get a working AI assistant with file access, code search, and shell commands
No vendor lock-in - Switch between OpenAI, Claude, DeepSeek, etc. anytime
Production-ready - Built-in session persistence, cost tracking, and error handling
Extensible - Add custom tools via MCP servers or local skill files
AI-aware - Context compression and specialized modes for different tasks

Built-in Tools

Tool	Function
`read`	Read files (text, images, PDF)
`write`	Write files
`edit`	Edit files (search & replace)
`glob`	Find files by pattern
`grep`	Search file contents
`bash`	Execute shell commands
`fetch`	Make HTTP requests
`task`	Launch specialized agents
`todo`	Track task progress
`askUserQuestion`	Interactive Q&A with users

📖 Tools Documentation

Use Cases

Build Custom AI Assistants - Chatbots, customer service systems
IDE Integration - Code assistants, smart autocomplete
Code Review - Automated code review systems
Debugging Tools - Intelligent error diagnosis
Education Platforms - Programming teaching assistants
Documentation Generation - Automated documentation generation

Documentation

Getting Started

5-Minute Quick Start - Get up to speed with OriCore
Tutorials - Practical examples

Core Features

API Reference - Complete API documentation
Configuration Guide - All configuration options
Tools System - Built-in tools guide
Interaction Modes - 5 professional modes

Advanced Features

Session Management - Persistence & context compression
Event System - Message bus & events
MCP Integration - MCP protocol support
Skill System - Custom skill loading
Approval System - Tool execution permission control

Acknowledgments

This project references the core architecture of the following excellent project:

neovate-code - Core AI engine architecture

OriCore has been refactored and streamlined on this foundation, removing UI, CLI, and other peripheral features to focus on providing a powerful and standalone AI engine library that can be easily integrated into any project.

License

MIT © lyw405

Support

GitHub: lyw405/oricore
Issues: Report a problem