RLM Batch — Parallel and Batched Execution

Run multiple independent tasks using Daytona sandboxes, or use rlm_query_batched for batched recursive sub-work inside a Daytona session.

There are no slash commands — all interactions use the Daytona Python API.

Concept

Batching has two shapes in daytona_pilot:

1. rlm_query_batched (inside sandbox code) — for concurrent recursive sub-queries from within a running Daytona session.
2. Multiple DaytonaInterpreter instances (host-side) — for independent parallel execution jobs on the host.