X-it: The Dual-Brain HR Assistant Agent

This is an advanced HR Assistant built on the A2A (Agent-to-Agent) compatible agent template. X-it serves two primary functions: conducting conversational exit interviews to generate detailed PDF summaries, and acting as an HR Data Analyst to generate aggregate attrition reports and charts. It implements the A2A JSON-RPC 2.0 protocol using FastAPI, giving you complete control over your stack.

Frameworks

• Server: FastAPI

• Agent Logic: LangChain (Pre-configured)

• Data & Reporting: Pandas, Matplotlib, FPDF2

Structure

• src/main.py: Protocol Implementation. A FastAPI server that handles the A2A JSON-RPC message/send requests.

• src/models.py: Data Models. A2A protocol Pydantic models (Message, Task, etc.).

• src/tools.py: Tools Definition. Contains the dual-tool logic (exit_interview_summary and generate_aggregate_attrition_report) for generating the dynamic HR PDFs.

• src/agent.py: Core Logic. Configures the LangChain agent and tools, featuring the customized "Dual-Brain" routing system prompt.

• src/attrition_data.csv: Dataset. The database tracking employee tenure, departments, reasons for leaving, and suggestions.

• AgentCard.json: Agent metadata (auto-generated).

• Dockerfile: Standard Python Dockerfile.

How to Use

1. Prepare your Data: Ensure attrition_data.csv contains your up-to-date employee records.

2. Customize (Optional): Edit src/tools.py or src/agent.py if you need to adjust PDF formatting or the LangChain agent's behavior.

3. Build and Run:

   docker build -t x-it-agent .
   docker run -p 5000:5000 -e OPENAI_API_KEY=your_key x-it-agent

Protocol Details

This template implements:

• POST /: JSON-RPC 2.0 endpoint.
  • Method: message/send
  • Params: A2A Message format