LLM Reasoning with Amazon Bedrock

🧠 Planner Skill

@skill(trigger="on_goal")
def bedrock_planner(context):
    response = run_reasoning(goal, memory=context.memory.recall_last_n(5))
    action = response.get("action")
    if action:
        context.agent.trigger(action)

Prompt Template Store

agent.templates.register("goal_plan", "You are solving: {goal}\nMemory: {memory}")

Fallback Logic

• Auto retries LLM if action missing
• Fallbacks to "noop" after max attempts