LangChain SigNoz MCP Demo

Using OpenInference and OpenTelemetry to send traces from your LangChain MCP agent app to Signoz

Getting Started

First, install all the necessary dependencies for the backend:

Optional Create Python virtual env:

python -m venv myenv && \
source myenv/bin/activate

Then:

pip install -r requirements.txt

Install all the necessary dependencies for the frontend:

cd frontend && \
npm install

Next create a .env file with the following(in root directory):

OPENAI_API_KEY=<your-openai-api-key>
SIGNOZ_INGESTION_KEY=<your-signoz-ingestion-key>

Run the fastapi backend:

uvicorn main:app --reload --port 8001

Run the frontend:

cd frontend && \
npm start

Setup MCP server: Follow the README in the following repo to setup and deploy the signoz mcp server locally (default localhost port 8000).

Open http://localhost:3000 with your browser to see the result and interact with the application.

After using the application, you should be able to view traces in your SigNoz Cloud platform:

Traces

You can also create custom dashboards using these traces and span attributes:

Import Dashboard

Go to the Dashboards tab in SigNoz.

Click on + New Dashboard

Go to Import JSON

Import the langchain-mcp-dashboard.json file from the repo.

Your dashboard should now be imported and look something like this: