Init research agent

[raychrisgdp]

Research Agent Implementation Checklist

Setup Phase

• Create directory structure for the Research Agent:

  • Create research_agent directory under aip_agent_quickstart/agents/
  • Create nested directory research_agent/research_agent/ for core logic
  • Create __init__.py files in all directories

• Create essential files:

  • Create pyproject.toml for dependency management
  • Create server.py (in outer research_agent directory)
  • Create Dockerfile for containerization
  • Create README.md with documentation
  • Create core agent files (in inner research_agent directory):
    • __init__.py
    • config.py for agent configuration
    • tools.py for custom tools (if needed)

Core Agent Implementation

• Implement core Research Agent logic:
  • Create LangGraphAgent implementation
  • Implement topic classification (academic vs travel)
  • Write system prompt/instruction that explains the agent's capabilities
  • Implement error handling for unsupported topics

MCP Server Integration

• Implement MCP server connection:
  • Import Arxiv MCP config from mcp_configs/configs.py
  • Add MCP server to agent using add_mcp_server
  • Test connection to Arxiv MCP server (note: may require VPN connection)

A2A Integration (Placeholder for future)

• Implement A2A agent integration:
  • Define A2A client configuration
  • Implement agent discovery
  • Register discovered agents with register_a2a_agents

Hello World Client Implementation

• Create hello world client for testing (hello_world_research_agent.py):
  • Create initial client that defines single agent
  • Add MCP server registration
  • Add A2A agent registration placeholders
  • Implement sample queries for testing

Testing Phase

• Test Research Agent with sample queries:

  Academic Queries (Should use Arxiv MCP tools)

  • "Find recent papers about quantum computing in cryptography"
  • "Summarize research trends in transformer models from 2023-2025"
  • "Search for papers by [Author Name] about [Topic]"

  Travel Queries (Should delegate to A2A agents when available)

  • "Plan a 3-day trip to Bali with budget considerations"
  • "What are popular tourist spots in Tokyo?"
  • "Find the best time to visit Paris and typical weather conditions"

  Unsupported Queries (Should respond with default message)

  • "What's the weather forecast for Jakarta today?"
  • "Write me a poem about mountains"
  • "Can you help me with my math homework?"

Deployment Phase

• Prepare for deployment:
  • Update docker-compose.yml to include Research Agent
  • Test containerized deployment with Podman:

    podman build -t research_agent .
    podman run -p 8002:8002 -e OPENAI_API_KEY=your-openai-api-key research_agent

  • Document deployment instructions in README.md

Documentation

• Complete documentation in README.md:
  • Document agent capabilities
  • Document query formats and examples
  • Document external service requirements (MCP, A2A)
  • Document expected response formats
  • Document Podman/Docker deployment instructions

Future Work

• Implement Web Search Agent integration
• Implement Information Compiler Agent integration
• Enhance topic classification with ML-based categorization
• Add more specialized tools for focused research tasks