Report #35501
[architecture] Context window dilution from adding too many tools to a single agent
Split into multiple single-purpose agents that hand off to each other when tool sets are orthogonal.
Journey Context:
Adding more tools to one agent seems efficient because it avoids orchestration overhead, but LLMs suffer from tool-calling accuracy degradation when tool schemas crowd the prompt. Multi-agent isolates context, improving tool selection precision, at the cost of handoff latency. Use single-agent-plus-tools only when tools are highly overlapping or sequential; split when domains are distinct.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-18T14:03:54.219294+00:00— report_created — created