shivonai-mcp

# tested on openai models and bedrock claude models # work with both types of models models. # ## Openai usage # import os # from llama_index.llms.openai import OpenAI # from llama_index.core.agent import ReActAgent # from shivonai.lyra import llamaindex_toolkit # # Set up OpenAI API key - you'll need this to use OpenAI models with LlamaIndex # os.environ["OPENAI_API_KEY"] = "openai_api_key" # # Your MCP server authentication details # MCP_AUTH_TOKEN = "shivonai_auth_token" # def main(): # """Test LlamaIndex integration with ShivonAI.""" # print("Testing LlamaIndex integration with ShivonAI...") # # Get LlamaIndex tools from your MCP server # tools = llamaindex_toolkit(MCP_AUTH_TOKEN) # print(f"Found {len(tools)} MCP tools for LlamaIndex:") # for name, tool in tools.items(): # print(f" - {name}: {tool.metadata.description[:60]}...") # # Create a LlamaIndex agent with these tools # llm = OpenAI(model="gpt-4") # # Convert tools dictionary to a list # tool_list = list(tools.values()) # # Create the ReAct agent # agent = ReActAgent.from_tools( # tools=tool_list, # llm=llm, # verbose=True # ) # # Test the agent with a simple query that should use one of your tools # # Replace this with a query that's relevant to your tools # query = "what listings I have?" # print("\nTesting agent with query:", query) # response = agent.chat(query) # print("\nAgent response:") # print(response) # if __name__ == "__main__": # main() # bedrock use """ Test script for LlamaIndex integration with ShivonAI package. This demonstrates how to use the llamaindex_toolkit to create and use tools with AWS Bedrock's Claude model. """ import os import boto3 from llama_index.llms.bedrock import Bedrock from llama_index.core.agent import ReActAgent from shivonai.lyra import llamaindex_toolkit # Your MCP server authentication details MCP_AUTH_TOKEN = "shivonai_auth_token" # Change this to your server URL def main(): """Test LlamaIndex integration with ShivonAI using AWS Bedrock's Claude model.""" print("Testing LlamaIndex integration with ShivonAI using Bedrock's Claude Sonnet...") # Get LlamaIndex tools from your MCP server tools = llamaindex_toolkit(MCP_AUTH_TOKEN) print(f"Found {len(tools)} MCP tools for LlamaIndex:") for name, tool in tools.items(): print(f" - {name}: {tool.metadata.description[:60]}...") # Create the Bedrock LLM with Claude Sonnet model llm = Bedrock( model="anthropic.claude-3-sonnet-20240229-v1:0", # Claude 3 Sonnet model ID # Authentication options - choose one: # Option 1: Using AWS profile # Option 2: Using AWS credentials aws_access_key_id="bedrock_access_key", aws_secret_access_key="bedrock_secrate_access_key", # aws_session_token="YOUR_SESSION_TOKEN", # if using temporary credentials region_name="bedrock_region", # Change to your AWS region # Required for newer Claude models context_size=200000, # Set appropriate context size for Claude 3 Sonnet # Optional parameters temperature=0.7, max_tokens=2000 ) # Convert tools dictionary to a list tool_list = list(tools.values()) # Create the ReAct agent agent = ReActAgent.from_tools( tools=tool_list, llm=llm, verbose=True ) # Test the agent with a simple query that should use one of your tools # Replace this with a query that's relevant to your tools query = "what listings I have?" print("\nTesting agent with query:", query) response = agent.chat(query) print("\nAgent response:") print(response) if __name__ == "__main__": main()