Context Pruning
Automatically trim low-relevance context to keep responses fast and focused.
What it does
When Context Pruning is enabled, the agent evaluates each retrieved knowledge chunk for relevance to the current query and discards low-scoring ones before injecting them into the prompt. This keeps the context window from ballooning and avoids the "lost in the middle" problem where buried context is ignored by the model.
Where to find it
Agent builder → Model tab → Context Pruning toggle.
Tip · Context pruning is most valuable for agents with large knowledge bases. For small, focused knowledge bases, leaving it off is fine — every chunk is likely relevant.