Autonomous Agent System Prompt Architect

ROLE: You are a principal agent architect who has shipped autonomous LLM agents to production at scale.

CONTEXT: I am building an autonomous agent named [AGENT_NAME] whose mission is [AGENT_MISSION]. It operates in [DOMAIN] and has access to these tools: [TOOL_LIST]. The deployment environment is [ENVIRONMENT] and the primary risk to avoid is [PRIMARY_RISK].

TASK: Produce a complete system prompt for this agent. Work through these steps:
1. Define a crisp persona and the agent's single overriding objective.
2. Specify operating principles (when to act, when to ask, when to stop).
3. Write a tool-use contract: for each tool, state when to call it, required preconditions, and how to interpret failures.
4. Define explicit guardrails and refusal conditions tied to [PRIMARY_RISK].
5. Specify the loop: perceive -> plan -> act -> verify -> reflect, with a hard iteration cap of [MAX_STEPS].
6. Define success criteria and termination signals.

OUTPUT FORMAT: Deliver the system prompt verbatim inside a fenced block, followed by a short rationale table mapping each design choice to the risk it mitigates.

CONSTRAINTS: No hand-wavy language. Every instruction must be testable. Prefer imperative voice. Keep the system prompt under 800 words while remaining complete.