InjoysAI LogoInjoysAI
Open Source · GitHub

Connect Multi-Channel IM with AI Agents,
Unified & Flexible

Universal multi-IM AI integration skeleton. Connect Telegram, WhatsApp, Discord and more to any AI Agent Runtime — with a clean, extensible 3-layer architecture.

View on GitHubContact Us
02
Interactive Demo

Simulated Pipeline Demo

Three-step simulated pipeline: message ingestion, adapter normalization, and agent runtime execution

Simulated Pipeline DemoInteractive Simulation
1. IM IngestionTelegram / WhatsApp / Discord
Ingest multi-channel webhooks & event streams, securely identifying user sessions
User sends to bot:"Summarize progress on today's tech tickets"
03
Decoupled Architecture

Decoupled 3-Tier Architecture

Ingestion, adapter, and execution tiers are fully decoupled, independently testable and extensible

3-Tier Architecture Overview · Open Source
01
Tier 1

IM Channel Ingestion Tier

Multi-Channel Ingestion

Supports protocol integrations and event listeners for Telegram, WhatsApp, and Discord, safeguarding network boundaries.

Webhook & Event Ingestion
02
Tier 2

Adapter & Routing Tier

Adapter & Session Routing

Normalizes multi-channel protocols into a unified JSON payload, orchestrating conversation state machines and routing.

Normalized Data Flow
03
Tier 3

Agent Runtime & Tools Tier

Pluggable Agent & MCP Tools

Adheres to standard Model Context Protocol (MCP) and Tools specifications, connecting to any pluggable AI Agent Runtime.

Decoupled Execution
04
Feature Matrix

Feature Matrix Bento Wall

Architected for real-world extensibility, clean abstractions, and rapid developer customization

IM Channels

Multi-IM channel support

Connect Telegram, WhatsApp, and Discord webhooks and event streams under a single unified standard.

Telegram · WhatsApp · Discord
Pluggable Runtime

Pluggable Agent Runtime

Decoupled from model vendors; easily switch between custom runtimes, LangChain, or proprietary engines.

LangChain / Custom Runtime
Standard Protocols

Standard MCP / Tools integration

Native compatibility with Model Context Protocol and function-calling schemas for seamless tool ecosystem adoption.

MCP & Tools Integrated
Production Ready

Production-ready skeleton architecture

Clean TypeScript strict typings, robust modular architecture, unified error handling, and structured logging.

TypeScript Clean Architecture
05
Specs & FAQ

Technical Specifications & FAQ

Fact-based recap of architecture and integration details without fabricated capabilities

Technical Specifications

Runtime Environment
Node.js 20+ / TypeScript
Supported Channels
Telegram / WhatsApp / Discord
Standard Protocols
MCP (Model Context Protocol) / REST
Open Source Repository
github.com/InJoysAI/im-agent-bridge
Q1: How does IM Agent Bridge connect to my AI Agent?
IM Agent Bridge provides standard MCP (Model Context Protocol) and Tools interfaces at the third tier, delivering normalized messages directly to any compatible Agent Runtime.
Q2: What message types are supported?
The adapter layer normalizes multi-channel text, events, and callback payloads into a unified JSON structure for downstream agent processing.
Q3: How is it deployed and run?
The project is a standard Node.js/TypeScript skeleton that can run standalone in Docker containers or serverless environments with your channel tokens.
Q4: Does it include pre-configured proprietary LLM models?
No. IM Agent Bridge focuses purely on channel ingestion, message normalization, and runtime bridging. Users bring their own models and runtimes.