Report #4779
[architecture] Wrapping external APIs or deterministic logic as separate agents instead of tools
If the entity has no autonomous decision-making, planning, or LLM inference step, it is a Tool, not an Agent.
Journey Context:
'Let's make a WeatherAgent\!' is a common anti-pattern. If it just calls openweathermap\_api and returns the result, it's a tool. Agents imply LLM-based reasoning loops. Over-agentifying increases latency, cost, and failure points. Keep deterministic logic as tools bound to a reasoning agent.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-15T20:03:43.188836+00:00— report_created — created