💡 Tools4AI

Tools4AI is 100% Java based Agentic AI Framework and light weight ADK library which can be used to build Java based AI agents for integration with enterprise Java applications. It can be used to build AI agents in A2A , MCP, A2UI, UCP and other protocols. This project illustrates the integration of AI Agents with enterprise tools or external tools, converting natural language prompts into agent actions. These prompts can be called "actionble prompts" or "agent prompts" By leveraging AI capabilities, it streamlines user interactions with complex systems, enhancing productivity and innovation across diverse applications.

For example , we can integrate AI Agent with a customer service application. Users can interact with the AI agent by asking questions or making requests in natural language. For example, a user might ask,"Schedule a maintenance appointment for my car." The AI agent interprets the request, extracts relevant information such as the service required and preferred date, and then triggers the appropriate agent action in the customer service application to schedule the appointment. This seamless integration streamlines the process for users and enhances the efficiency of the customer service workflow through agentic ai automation.

Prompt	Action
Create a new task for the marketing campaign.	The AI agent interprets the request and generates a new task entry within the project management tool dedicated to the marketing campaign, assigning it relevant details such as priority level, due date, and task description.
Generate a sales report for the previous quarter.	The AI agent accesses data from the company's sales database, analyzes the information for the previous quarter, and generates a comprehensive sales report, which is then delivered to the user or stored in the appropriate location for access.
Check the inventory status of product X.	The AI agent retrieves real-time inventory data for product X from the inventory management system and provides the user with information regarding current stock levels, including quantities available, locations, and any pending orders.
Schedule a video conference with the engineering team for next Monday at 10 AM.	The AI agent interfaces with the calendar and scheduling tool, creates a new event titled "Engineering Team Video Conference" for the specified date and time, and sends out meeting invitations to all members of the engineering team.
Submit a reimbursement request for the business trip expenses.	The AI agent guides the user through the reimbursement request process, collecting necessary details such as expense receipts, dates, amounts, and purpose of expenditure. Once compiled, the system submits the reimbursement request to the appropriate department for processing.

Prompt prediction is a technique used to anticipate user actions based on their input prompts. For instance, if a user's prompt is "my car broke down," in addition to the action "bookTaxi," the AI agent can predict a set of subsequent actions such as "bookCarService" and "orderFood" (if it's dinner time). This predictive capability enhances the user experience by proactively suggesting relevant actions or services based on the context provided in the prompt.

Table of Contents

• Rapid Start
• Tools4AI
• Architecture & Comparison with Spring AI
• SetUp
  • Gemini
  • OpenAI
  • Anthropic
• Reference Examples
  • Java Actions
  • HTTP REST Actions
  • Shell Actions
  • Custom HTTP Actions
  • Selenium Integration
  • Image Actions
  • Automated UI validation
  • Spring Integration
  • JavaDocs
• Prediction Loaders
  • Java Prediction Loader
  • Swagger Prediction Loader
  • Shell Prediction Loader
  • Extended Prediction Loader
• Agent Toolkit
  • Memory
  • Multi-Agent Orchestration
  • ReAct Planning

📌 Rapid Start

🧱 Do you want to start building ASAP , Look at Rapid start here https://github.com/vishalmysore/agenticjava

🌱 Integration of Spring Controller and AI Actions - https://github.com/vishalmysore/SpringActions

🧭 Architecture & Comparison with Spring AI

Tools4AI is a retrofit layer — it makes any existing Java system AI-controllable with minimal code change.
Spring AI is a platform for building AI-first apps from scratch on Spring Boot.

Dimension	Tools4AI	Spring AI
Core purpose	Agentic action routing — maps prompts to pre-existing Java methods/REST/shell	AI primitives — chat, embeddings, RAG, vector stores for Spring apps
Entry point	Annotate any existing Java class/method — works without Spring	Built on Spring Boot — Spring context is required
Action discovery	Annotation-driven classpath scan + YAML/JSON config for REST/shell	Manual @Tool registration, function callbacks
Parameter mapping	Fully automatic — POJOs, Lists, Maps, arrays populated from prompt	Manual — you define function schemas explicitly
Non-Java actions	First-class shell scripts, Swagger/OpenAPI, HTTP REST — no code needed	Not supported natively
Safety layer	GuardRails, HumanInLoop, hallucination/bias/fact detectors	Not built in
RAG / Embeddings	Not present	Core feature
Weight	Lightweight, single JAR	Full Spring ecosystem

📄 Full architecture deep-dive, capability breakdown, and improvement roadmap → ARCHITECTURE.md

🤖 Agent Toolkit

The com.t4a.agent package adds composable agentic building blocks on top of any AIProcessor (OpenAI, Gemini, Anthropic, LocalAI). Everything is a plain decorator or helper — no existing code changes, no Spring required, and the pieces compose freely.

Memory / Conversation History

Give your agent short-term (sliding window) or long-term (file-backed, survives restarts) memory:

AgentMemory memory = new InMemoryAgentMemory(20);                               // last 20 turns
AgentMemory memory = new PersistentFileAgentMemory("/var/agent/session.json");  // durable

memory.addTurn("What is the weather?", "25°C");
String prompt = memory.getHistoryAsContext() + "\nNow: what should I wear?";

Multi-Agent Orchestration

Register multiple specialised agents and let the LLM route each part of a compound prompt to the right one, then synthesise a combined answer:

AgentOrchestrator orch = new AgentOrchestrator(new OpenAiActionProcessor());
orch.register(new AgentDefinition("flights", "handles flight booking", flightProcessor));
orch.register(new AgentDefinition("hotels",  "handles hotel reservations", hotelProcessor));

OrchestrationResult result = orch.execute(
    "Book a flight to Bangalore on Aug 15 and a hotel for 3 nights");
System.out.println(result.getSummary());

Re