networkAI-A2A

Agent-to-Agent (A2A) AI System for Natural Language to Elasticsearch Querying and Visualization

network-A2A is an experimental modular AI system where multiple autonomous agents collaborate to answer natural language queries over Elasticsearch datasets. Inspired by multi-agent architectures and Model Context Protocol (MCP), each agent has a distinct role: planning, data retrieval, and visualization.

---

🧠 Key Concepts

• PlannerAgent: Interprets user questions and decomposes them into tasks.
• ElasticAgent: Generates and executes Elasticsearch queries using LLM + schema context.
• VizAgent: Visualizes structured results via graphs or tabular summaries.
• Messaging Layer: Standardized message format supports signatures (SACMP-ready).
• MCP Compatibility: Agents can be MCP-compliant tools or communicate in A2A mode.

---

📂 Project Structure

networkAI-a2a/
├── agents/
│   ├── planner_agent.py        # Understands intent and plans actions
│   ├── elastic_agent.py        # Generates and executes ES queries
│   └── viz_agent.py            # Handles visualization
│
├── messaging/
│   ├── base_agent.py           # Abstract base for all agents
│   ├── message.py              # Message format (incl. SACMP-ready signatures)
│   └── transport.py            # Future: handles async queue or bus
│
├── memory/
│   └── registry.py             # (Optional) memory or long-term storage
│
├── prompt_1.txt                # System prompt template (injects schema)
├── user_interface.py           # CLI or future API for user interaction
├── main.py                     # Manual A2A orchestrator chain
├── .env                        # OpenAI API Key and env config

---

🚀 Quick Start

1. Clone and setup

git clone https://github.com/your-org/networkAI-a2a.git
cd ntsctAI-a2a

2. Environment setup

Install dependencies:

pip install -r requirements.txt

Create .env:

OPENAI_API_KEY=your_openai_key_here

3. Run the demo

python main.py

You’ll be prompted for a natural language query. The system runs the following chain:

 ------         --------------        --------------        -----------         ---------------
| User |   →   | PlannerAgent |   →  | ElasticAgent |   →  |  VizAgent |   →   | Chart Display |
 ------         --------------        --------------        -----------         ---------------

---

🧩 Example Queries

• What are the top 5 servers with the highest packet loss?
• Which IP has the most unique applications served?
• List the top 3 clients generating DNS traffic to 8.8.8.8.

---

🔐 SACMP Compatibility (Planned) • All messages support digital signatures per agent. • Memory entries (e.g., vector results, plans) will be agent-signed. • Replay protection, schema fingerprinting, and provenance are in scope.

---

🧱 Extending

Component You can add… ElasticAgent Support for multiple backends (Splunk, Loki) VizAgent Graph export, summary tables, HTML reports Message signature, trace_id, or auth_context fields Transport Real A2A: Redis, MQTT, or REST messaging bus

---

🤝 License & Credits

• MIT License
• Inspired by ideas from MCP (Model Context Protocol) and A2A architectures.

---

📬 Contact

• Maintainer: [fromzy]
• Mail: [fromzy at gmail dot com]
• Project status: Research-grade, evolving toward production.