Durable storage backend support (NATS JetStream / Redis / Postgres)?

Hi team 👋

I’ve been evaluating a2a-go for use in a multi-VM environment (Fly.io single-container VMs).

One limitation I’ve run into is that the current task manager and event queue appear to be in-memory only, meaning:

- tasks and events are not durable
- state is lost on process/VM restart
- multi-node / cluster deployments can’t coordinate work

Before building my own backend layer, I wanted to ask:
Is there any existing or planned support for a durable backend?

(E.g. NATS JetStream, Redis, or Postgres)
Specifically, I’m looking for a way to persist:
task queue
task execution state
event log
task results

The Python ecosystem has a2a-redis, but I couldn’t find an equivalent for Go.
Is durable storage on the roadmap for a2a-go, or should implementers provide their own backend via the extension points being discussed in issue #115?

Thanks!

The issue mentions extension points from issue #115; start by reading that discussion to understand the backend interface. Look at the task manager and event queue code to see where storage is currently in-memory. Check if there are existing examples of backend implementations in the Python ecosystem's a2a-redis. 'Done' means a durable backend for task queue, execution state, event log, and results using NATS JetStream, Redis, or Postgres.

Scritto dal modello di indicizzazione a partire dal testo della issue.