Superagentx - OpenSource Agent AI with Bedrock LLMs initial examples

agents-and-function-calling/open-source-agents/superagentx/01_superagentx_simple_bedrock_get_weather_tutorial.ipynb

@@ -0,0 +1,323 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "9f3de924-5afd-4df3-b81b-1d2950390a29",
+   "metadata": {},
+   "source": [
+    "# Simplified AWS Bedrock LLM function calling using SuperAgentX.\n",
+    "\n",
+    "### Open Source Github Repository URL: https://github.com/superagentxai/superagentx\n",
+    "\n",
+    "## What is SuperAgentX?\n",
+    "\n",
+    "SuperAgentX is an advanced agentic AI framework designed to accelerate the development of Artificial General Intelligence (AGI). It provides a powerful, modular, and flexible platform for building autonomous AI agents capable of executing complex tasks with minimal human intervention.\n",
+    "\n",
+    "Using the SuperAgentX agentic AI framework, function calling with AWS Bedrock LLMs is simplified.\n",
+    "\n",
+    "The below example creates weather as a handler (Tool) and invokes using SuperAgentX's agent.\n",
+    "\n",
+    "#### The Below example explains how to invoke Bedrock Converse API using the SuperAgentX framework.\n",
+    "\n",
+    "In this example, the SuperAgentX Agent module can automatically invoke a mock weather handler (Tool).\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "36e84a3e-07c5-4bfe-a00c-1e95ac06be23",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!pip install superagentx"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "id": "49046b9a-5b97-4820-854c-2fd47da4c5d0",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from superagentx.agent import Agent\n",
+    "from superagentx.engine import Engine\n",
+    "from superagentx.llm import LLMClient\n",
+    "from superagentx.handler.base import BaseHandler\n",
+    "from superagentx.prompt import PromptTemplate\n",
+    "from superagentx.agentxpipe import AgentXPipe"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "id": "526ddbe1",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<style>\n",
+       "table {float:left}\n",
+       "</style>\n"
+      ],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    }
+   ],
+   "source": [
+    "%%html\n",
+    "<style>\n",
+    "table {float:left}\n",
+    "</style>"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "e3973645-f92f-44b6-82be-e76e994a5386",
+   "metadata": {},
+   "source": [
+    "Although this example leverages Claude 3 Sonnet, Bedrock supports many other models. The full list of models and supported features can be found [here](https://docs.aws.amazon.com/bedrock/latest/userguide/conversation-inference.html). The models are invoked via `bedrock-runtime`.\n",
+    "\n",
+    "\n",
+    "**Best Practice**: Always set your AWS credentials as environment variables.\n",
+    "\n",
+    "```\n",
+    "export AWS_ACCESS_KEY=<<`YOUR_ACCESS_KEY`>>\n",
+    "export AWS_SECRET_KEY=<<`YOUR_ACCESS_SECRET_KEY`>>\n",
+    "export AWS_REGION=<<`AWS_REGION`>>\n",
+    "```\n",
+    "\n",
+    "#### LLM Config in SuperAgentX\n",
+    "\n",
+    "LLM Configuration Specify ==> llm_config\n",
+    "\n",
+    "| Name | Description | Data Type | Required | Example |\n",
+    "|------|-------------|-----------|----------|---------|\n",
+    "|`model`| AWS Bedrock supported [models](https://docs.aws.amazon.com/bedrock/latest/userguide/conversation-inference.html)|str| Yes| 'model': 'anthropic.claude-3-5-sonnet-20240620-v1:0' |\n",
+    "|`llm_type`| LLM type - `bedrock`|str| Yes| 'llm_type':'bedrock'\n",
+    "\n",
+    "\n",
+    " "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "id": "64768dc6-5f0f-4d26-b0ca-f40f63834fb1",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "llm_config = {'model': 'anthropic.claude-3-5-sonnet-20240620-v1:0', 'llm_type':'bedrock'}\n",
+    "\n",
+    "llm_client: LLMClient = LLMClient(llm_config=llm_config)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "c3d3ed52-5ff9-4c2c-ad7a-42fd8f453fa3",
+   "metadata": {},
+   "source": [
+    "### Handler (Tool) definition \n",
+    "\n",
+    "We define `WeatherHandler` as a class where individual handlers (tools) are defined as functions. Note that there is nothing specific to the model used or Bedrock in this definition."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "id": "26ebfd29-1069-48a5-854a-9f6af856f9a5",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from superagentx.handler.base import BaseHandler\n",
+    "from superagentx.handler.decorators import tool\n",
+    "\n",
+    "class WeatherHandler(BaseHandler):\n",
+    "    \n",
+    "    @tool\n",
+    "    async def get_weather(self, city: str, state: str) -> str:\n",
+    "        \"\"\"\n",
+    "        Generate weather in Celsius or Fahrenheit \n",
+    "\n",
+    "        Args:\n",
+    "            @param city (str): The city name\n",
+    "            @param state (str): The state name\n",
+    "\n",
+    "            @return result (Str): Return Fake Weather for the given city & name in Fahrenheit  \n",
+    "        \"\"\"\n",
+    "        result = f'Weather in {city}, {state} is 70F and clear skies.'\n",
+    "        print(f'Tool result: {result}')\n",
+    "        return result"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "79a5dd9f-d43e-4f8d-a144-a186f1af30d3",
+   "metadata": {},
+   "source": [
+    "### Create Prompt Template Object"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "id": "3a732b3a-e562-4703-854e-f1ab3a0595a9",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "prompt_template = PromptTemplate()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "id": "adf0cc3f-50c2-4ddf-b97d-967ba5222731",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "weather_forecast_engine = Engine(\n",
+    "    handler=WeatherHandler(),\n",
+    "    llm=llm_client,\n",
+    "    prompt_template=prompt_template\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "bd7a5bae",
+   "metadata": {},
+   "source": [
+    "## Agent attributes\n",
+    "\n",
+    "| Attribute                  | Parameter  | Description                                                                                                                                                                                                                                    |\n",
+    "| :------------------------- | :--------- | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n",
+    "| **Role**                   | `role`     | Defines the agent's function within the SuperAgentX. It determines the kind of activcity the agent is best suited for.                                                                                                                                    |\n",
+    "| **Goal**                   | `goal`     | The individual objective that the agent aims to achieve. It guides the agent's decision-making process.                                                                                                                                        |\n",
+    "| **LLM**              | `llm`| Represents the language model that will run the agent.|\n",
+    "| **Retry Max** *(optional)*       | `max_retry`      | sets the maximum number of iterations an agent can complete before it must provide its best possible answer. Default `5`                                                       |\n",
+    "| **Prompt Template**       | `prompt_template`      | Specifies the prompt format for the agent. |                                                       |\n",
+    "| **Engine**    | `engines`      | A list of engines (or lists of engines) that the engine can utilize in `parallel` or `sequnetial`. This allows for flexibility in processing and task execution based on different capabilities or configurations or or configurations.|"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "id": "31b003a7-b8de-44c7-bd73-5900a4f84a03",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "weather_man_agent = Agent(\n",
+    "    goal=\"Verify the weather results for the given city and state \",\n",
+    "    role=\"Weather Man\",\n",
+    "    llm=llm_client,\n",
+    "    max_retry=2, #Default Max Retry is 5\n",
+    "    prompt_template=prompt_template,\n",
+    "    engines=[weather_forecast_engine],\n",
+    "        )"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "04315d06",
+   "metadata": {},
+   "source": [
+    "### Pipe Attributes\n",
+    "\n",
+    "| Attribute                  | Parameter  | Description                                                                                                                                                                                                                                    |\n",
+    "| :------------------------- | :--------- | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n",
+    "| **Name**                   | `name`     | An optional name for the agentxpipe, providing a more friendly reference for display or logging purposes.                                                                                                                                    |\n",
+    "| **Agents**                   | `agents`     | A list of Agent instances (or lists of Agent instances) that are part of this structure. These agents can perform tasks and contribute to achieving the defined goal.         |\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "id": "daa61f68",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "pipe = AgentXPipe(\n",
+    "            name='Fake Weather Man',\n",
+    "            agents=[weather_man_agent]\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "id": "961a636e-c8ef-4142-bc83-5a8ea086374e",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Tool result: Weather in New York, New York is 70F and clear skies.\n"
+     ]
+    }
+   ],
+   "source": [
+    "result = await pipe.flow(query_instruction=\"What is the weather like in New York, USA?\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "id": "82b00c28-ea7c-4817-a1b9-0babf15c2fd0",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "[GoalResult(name='Agent-9c51452a31fd417eb6492a01961e7011', agent_id='9c51452a31fd417eb6492a01961e7011', reason='The output context provides weather information for New York, USA, which matches the query instruction.', result='Weather in New York, New York is 70F and clear skies.', content=None, error=None, is_goal_satisfied=True)]\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(result)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "fd2dec9d-0037-4298-abcb-58b5a309eade",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "a2640345",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "venv",
+   "language": "python",
+   "name": "venv"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.12.5"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}