prisma.fleet.run_pool owns the multi-host thread-pool loop that screen
had inline: at most --per-host in-flight requests per Ollama host, LLM
calls only on the pool, every SQLite touch and per-item commit on the
calling thread, and a host that fails twice in a row is retired for the
rest of the run (its queued items go to the surviving hosts). screen,
eligibility and extract each supply prepare/apply/on_error glue; items
with no retrievable full text resolve without a fleet call. CLI:
prisma eligible/extract gain --hosts/--fleet-model/--num-ctx/--per-host
matching prisma screen. Docs regenerated.
49a2f6f723