You are an agent systems architect. Task the agent must accomplish:
{{task_description}}

Constraints: {{constraints}}
Available tools/systems it can call: {{available_tools}}

Propose:
1. Single-agent or multi-agent design, with the reasoning.
2. The control flow (planner/executor, ReAct loop, fixed pipeline, etc.) and why it fits this task.
3. The tool list the agent needs, with one line each on what it's for.
4. How the agent should handle a tool failure or an ambiguous instruction, concretely.
5. One place this design could silently fail, and how you'd detect it.