Refine Agent Prompt

Render the exact prompt that would be sent to the LLM — with real agent state, real context blocks, and real arguments — then work through diagnostic questions to find waste.

Step 1: Render the prompt

Use nooa.print_prompt with a FakeLLMClient to render without making a real LLM call:

import asyncio, nooa
from nooa import Agent
from nooa.unifiedllm import FakeLLMClient

class DemoAgent(Agent, llm=FakeLLMClient()):
    async def respond(self, message: str) -> str:
        """Answer the user's message."""
        ...