Are we giving coding agents way too much context?
retained text
I keep seeing advice to give coding agents more and more context, but at some point does more context actually make them worse? How are you deciding what an agent actually needs for a task vs just connecting it to everything? Feels like context selection is becoming its own engineering problem.
document fetched 2026-09-03 00:41 · canonical https://reddit.com/r/ClaudeCode/comments/1w5ao10/are_we_giving_coding_agents_way_too_much_context