How to actually use the API after running autogenstudio serve?

[sedatepace]

💬 Title

How to actually use the API after running autogenstudio serve?

---

🧩 Body

Hi everyone 👋

I successfully ran:

autogenstudio serve --team RoundRobin_OneShot.json --host 127.0.0.1 --port 8084

The server starts fine, and I can open the /docs endpoint in the browser to see the OpenAPI UI.
However, the documentation doesn’t clearly explain how to actually send a request to the running API.

I tried this command:

curl -X POST "http://127.0.0.1:8084/predict" \
  -H "Content-Type: application/json" \
  -d '{"task": "default", "input": "1+1은?"}'

but the response was:

{"detail":"Not Found"}

So I’m wondering:

1. What are the correct endpoints to use after running serve?
2. What should the JSON payload look like (for example: { "task": "...", "input": "..." })?
3. Is there an example of how to interact with a custom team (like RoundRobin_OneShot.json) through the served API?

Would really appreciate an example or a short “how to test your served workflow via curl / Postman” guide 🙏

Environment

• OS: Ubuntu 22.04
• Python: 3.11
• Autogen Studio version: (please fill in, e.g. 0.1.7)
• Command used: autogenstudio serve --team RoundRobin_OneShot.json --port 8084

Thanks in advance!

[Shashank0701-byte]

You're using the wrong endpoint. The correct endpoint is /api/chat, and the payload needs to specify your workflow ID (the filename).
Here is the correct curl command

curl -X POST "http://127.0.0.1:8084/api/chat"
-H "Content-Type: application/json"
-d '{
"workflow": "RoundRobin_OneShot",
"message": "1+1은?"
}'

I hope this helps