is there a plan for persisting agent memory

[mstrYoda]

Currently agents have an in-memory memory to use in steps.
It would be nice to have persistent agent memory for tasks, so repetitive tasks will get the final answer quickly.

[aymeric-roucher]

For this, you could save the agent logs (by just copying attribute agent.logs) as your memory, do the modifications that you need, then manually modify another agent's logs to incorporate past memory! If you have specific issues where you think some building blocks would help, don't hesitate to provide an example and I'll work on it!