Introduction

Welcome to AgentForce ADK

Version: 0.12.0 (beta release)

TypeScript Cross-Runtime AI Agents AI Workflows

AgentForce ADK is a powerful TypeScript framework designed for creating, managing, and orchestrating AI Agent Workflows. Built with easy to use TypeScript classes and functions, it provides a simple yet powerful interface to develop Agent-Based applications with support for multiple AI providers and models.

What is AgentForce ADK?

AgentForce ADK (Agent Development Kit) simplifies the process of building AI-powered applications by providing:

Fluent Interface: Method chaining for intuitive agent configuration
Multi-Provider Support: Works with local (Ollama) and cloud (OpenRouter) providers
Type Safety: Full TypeScript support with comprehensive type definitions
Cross-Runtime: Seamless operation across Bun, Node.js, and Deno
Server Capabilities: Built-in HTTP server functionality for API deployment
Extensive Documentation: Guides, examples, and API references to help you get started quickly

Key Concepts

Agents

Agents are the Core building blocks of AgentForce ADK. Each agent represents an AI-powered entity that can:

Process prompts and generate responses
Be configured with different AI providers and models
Be deployed as HTTP endpoints
Chain operations using a fluent interface
Agents with skill injection can perform complex tasks by combining multiple skills and capabilities.
ToolUse allows agents to interact with external systems and APIs, enhancing their capabilities.

Providers

Providers are the bridge between your agents and AI services:

Ollama: Local model execution for privacy and development
OpenRouter: Cloud-based access to multiple AI providers (OpenAI, Anthropic, Google, Meta)
Native Providers: Direct integration with OpenAI, Anthropic, Google (coming soon)

Workflows

Workflows orchestrate multiple AI agents to work together on complex tasks:

Multi-Agent Coordination: Manage and coordinate multiple agents within a single workflow
Execution Patterns: Support for sequential, parallel, and iterative agent execution
Shared State Management: Store and share data between agents using shared storage
Dispatcher Agent: Use a coordinator agent to manage workflow execution and decision-making
Error Handling: Built-in success and failure handlers for robust workflow execution
Flexible Orchestration: Chain, loop, and iterate through complex multi-step processes

Server

Server capabilities enable deploying agents and workflows as HTTP web services:

HTTP Server Framework: Built on Hono framework for lightweight, fast HTTP handling
Agent-Powered Routes: Serve agents as API endpoints with optional schema validation
Static Routes: Add predefined routes that return static data for health checks and configuration
Form Triggers: HTML form integration that executes agents on form submission
Workflow Triggers: Execute workflow files through HTTP endpoints
API Compatibility: OpenAI and Ollama compatible endpoints for drop-in replacement
Multi-Service Deployment: Host multiple agents and workflows on a single server instance

Method Chaining

AgentForce ADK uses a Fluent Interface pattern that allows you to chain configuration methods:

const agent = new AgentForceAgent(config)
  .useLLM("ollama", "gemma3:12b")
  .systemPrompt("You are a helpful assistant")
  .prompt("Hello, how can I help?")
  .output("json");

Why Choose AgentForce ADK?

Developer Experience

Minimal Boilerplate: Create powerful agents with just a few lines of code
IntelliSense Support: Full IDE support with TypeScript autocompletion
Comprehensive Testing: Well-tested framework with extensive test coverage
Clear Documentation: Detailed guides and API references

Flexibility

Multiple Runtimes: Works with Bun, Node.js, and Deno
Provider Agnostic: Switch between local and cloud providers easily
Output Formats: Support for Text, JSON/YAML, and Markdown outputs
Deployment Options: Use as library or deploy as HTTP service

Bejond Hallo World

Advanced Features: Support for multi-agent workflows, parallel execution, and more
OpenAI Compatibility: Use AgentForceServer with OpenAI API compatible endpoints
Structured Logging: Built-in logging with structured output
Error Handling: Comprehensive error handling and validation
Performance: Optimized for both development and production use

What’s Next?

Now that you understand the basics, let’s get AgentForce ADK installed and create your first agent:

Installation Guide - Set up AgentForce ADK in your project
Provider Setup - Configure AI providers (Ollama, OpenRouter)
Quick Start - Create your first agent
Server Mode - Deploy agents as HTTP APIs

Ready to build your first AI agent? Let’s continue with the installation guide!