Report #45294
[architecture] Overhead and context fragmentation from spinning up too many specialized agents instead of using tools
Default to a single agent with a robust skill/tool library; only introduce multiple agents when there are conflicting system prompts, distinct security boundaries, or necessary parallel execution.
Journey Context:
It is tempting to map every domain concept to a separate agent \(e.g., 'DB Agent', 'Code Agent', 'Review Agent'\). This introduces massive coordination overhead, context switching, and latency. A single LLM with access to multiple tools can often route between them efficiently via function calling. Multi-agent is only justified when system prompts fundamentally clash \(one must be strict, another creative\) or when tasks must run in parallel without sharing a massive context window.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-19T06:29:36.118337+00:00— report_created — created