CCA Prep Q11 / 60 Context Management & Reliability
2:00:00
Claude Code for Continuous Integration

Your CI pipeline uses a multi-agent Claude Code setup where a parent orchestrator spawns subagents to run parallel test analysis across 12 service modules. After scaling from 6 to 12 services last sprint, you begin seeing "Prompt is too long" errors in the subagents' first turns — before any analysis has been completed. Examining the parent session configuration, you see it has 9 MCP servers registered, of which only 2 are relevant to the test analysis workflow. What change most directly eliminates the subagents' context window errors?