Follow these steps to run the MCP Pinecone tool and chatbot inside a Meshagent Room in the cloud for quick testing. When the room closes (e.g., due to inactivity), the MCP server and chatbot are removed automatically.

​
Environment Variables

  • PINECONE_ASSISTANT_HOST: Your Pinecone Assistant host.
  • PINECONE_API_KEY: Your Pinecone API key.

​
Steps

  1. Install Meshagent: 
    pip install "meshagent[all]"
  2. Sign Up & Authenticate: Follow the Meshagent CLI getting started guide.
  3. Start Pinecone MCP Service in a Room: 
    meshagent service test \
  --room=test \
  --role=agent \
  --image=meshagent/mcp-pinecone:latest \
  --env MESHAGENT_PORT=8001 \
  --env PINECONE_ASSISTANT_HOST=<YOUR_PINECONE_ASSISTANT_HOST_HERE> \
  --env PINECONE_API_KEY=<YOUR_PINECONE_API_KEY_HERE> \
  --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" \
  --name=mcp-pinecone-service-test
    This command will start a Meshagent Room named “test”, running the Pinecone MCP server. The room and its services will close if inactive.
  4. Join a Chatbot with the Toolkit in the Room: 
    meshagent chatbot join \
  --room=test \
  --agent-name=mcp-pinecone \
  --name=mcp-pinecone \
  --toolkit=mcp-pinecone
    • The chatbot will be able to use Pinecone tools in the room.
    • Room link will be printed in the output for browser-based chat.
  5. Test the Integration:
    • Open the room link from the chatbot join command.
    • Send a message to the agent to interact live with the MCP Server tools.

​
Project Level Deployment

To make the MCP server tools and a chatbot persistently available in every room of a Meshagent project (ideal for production), deploy as project-level services:
  1. Create Persistent MCP Pinecone Service: 
    meshagent service create \
  --role=agent \
  --image=meshagent/mcp-pinecone:latest \
  --env MESHAGENT_PORT=8001 \
  --env PINECONE_ASSISTANT_HOST=<YOUR_PINECONE_ASSISTANT_HOST_HERE> \
  --env PINECONE_API_KEY=<YOUR_PINECONE_API_KEY_HERE> \
  --port="num=8001 path=/webhook liveness=/ type=meshagent.callable" \
  --name=mcp-pinecone-service
  2. Create Persistent Chatbot Service with Toolkit: 
    meshagent service create \
  --image="meshagent/cli:latest" \
  --port="num=9001 path=/agent liveness=/ type=meshagent.callable participant_name=mcp-pinecone-chatbot" \
  --env="MESHAGENT_PORT=9001" \
  --name="mcp-pinecone-chatbot-service" \
  --command="meshagent chatbot service --agent-name=mcp-pinecone-chatbot --toolkit=mcp-pinecone"
These services will join every new room created within your Meshagent project, so your tools and chatbot are always available—no need to run them manually!

​
Tools Available

​
assistant_context

Retrieves relevant document snippets from your Pinecone Assistant knowledge base. Short Description:
assistant_context | Retrieves relevant document snippets from your Pinecone Assistant knowledge base.

​
Tool Details

assistant_context
Retrieves relevant document snippets from your Pinecone Assistant knowledge base.
Returns an array of text snippets from the most relevant documents.
  • assistant_name (string): Name of an existing Pinecone assistant
  • query (string): The query to retrieve context for.
  • top_k (integer, optional): Number of context snippets to retrieve. Defaults to 15. Recommended: 5-8 for simple/narrow queries, 10-20 for complex/broad topics.

​
Meshagent & MCP Resources

For more details about MCP servers and Pinecone’s Assistant integration, see Pinecone Assistant MCP Server Repository and this Docker image.