Content
# multiple-agents-on-single-a2a-server
This repository demonstrates how to run **multiple A2A agents** on the **same host** using the A2A protocol.
Each agent is served at a **unique URL path**, making it possible to host different agents without requiring multiple servers or ports.
---
## 📌 Example Setup
Three agents running on the same host:
| Agent Name | Agent card URL |
|-----------------------|-------------------------------------------------------------------------------------------------------------------|
| Conversational Agent | [http://localhost:8000/a2a/conversation/agent-card.json](http://localhost:8000/a2a/conversation/agent-card.json) |
| Trending topics Agent | [http://localhost:8000/a2a/trending/agent-card.json](http://localhost:8000/a2a/trending/agent-card.json) |
| Analyzer Agent | [http://localhost:8000/a2a/analyzer/agent-card.json](http://localhost:8000/a2a/analyzer/agent-card.json) |
---
## 🚀 Running Agents Locally
1. Clone the repository
```bash
git clone https://github.com/satendrakumar/multiple-agents-on-single-a2a-server.git
cd multiple-agents-on-single-a2a-server
```
2. Install dependencies (using uv)
```bash
uv sync
```
3. Set environment variables
* Copy `.env-sample` to `.env`
```bash
cp .env-sample .env
```
* Update values as needed
4. Start the agents
```bash
uv run main.py
```
---
## 🧪 Testing Agents using A2AClient:
Run the provided client app to send test requests:
```bash
uv run a2a_client_app.py
```
## Testing Agents Using CURL request:
```bash
Request:
curl --location 'http://localhost:8000/a2a/conversation/' \
--header 'Content-Type: application/json' \
--header 'Accept: text/event-stream' \
--data '{
"id": "6a39c736-fff7-45f8-b2b0-c44e705d2474",
"jsonrpc": "2.0",
"method": "message/stream",
"params": {
"configuration": {
"acceptedOutputModes": [],
"blocking": "True"
},
"message": {
"contextId": "8bffff7b-3abc-4d85-bcce-0bcdbe321017",
"kind": "message",
"messageId": "82642fd2-f270-4a56-a7d9-5d26fddabc95",
"parts": [
{
"kind": "text",
"text": "Who is PM of India?"
}
],
"role": "user"
}
}
}'
Reponse:
{
"id": "6a39c736-fff7-45f8-b2b0-c44e705d2474",
"jsonrpc": "2.0",
"result": {
"artifact": {
"artifactId": "9474e1ac-1e35-425c-9c18-238b07739e6d",
"parts": [
{
"kind": "text",
"text": "As of today, August 16, 2025, the Prime Minister of India is **Narendra Modi**.\n\nHe has been serving as the Prime Minister since May 26, 2014, and was sworn in for his third consecutive term on June 9, 2024, following the 2024 Parliamentary elections."
}
]
},
"contextId": "8bffff7b-3abc-4d85-bcce-0bcdbe321017",
"kind": "artifact-update",
"lastChunk": true,
"taskId": "23a93002-686b-4c8a-8bb6-ec649acea0c3"
}
}
```
---
## How to run inside Docker:
```bash
# Docker build
docker build -t multiple-agents-on-single-a2a-server:v1.0.0 .
# Docker run
docker run --env-file ./.env -e PORT=8000 -p 8000:8000 multiple-agents-on-single-a2a-server:v1.0.0
```
---
## Notes
- This setup demonstrates hosting multiple agents via unique URL paths behind a single application server.
- If you run behind a reverse proxy, ensure the `/a2a/...` paths are forwarded to the app.
- For local development, keep your working directory at the project root so relative imports and paths resolve correctly.
