traceAI-mongodb

OpenTelemetry instrumentation for MongoDB Atlas Vector Search.

Installation

pip install traceAI-mongodb

Quick Start

from fi_instrumentation import register
from traceai_mongodb import MongoDBInstrumentor
from pymongo import MongoClient

trace_provider = register(project_name="my-rag-app")
MongoDBInstrumentor().instrument(tracer_provider=trace_provider)

client = MongoClient("mongodb+srv://...")
db = client["mydb"]
collection = db["documents"]

# Vector search operations are traced
results = collection.aggregate([
    {
        "$vectorSearch": {
            "index": "vector_index",
            "path": "embedding",
            "queryVector": [0.1, 0.2, ...],
            "numCandidates": 100,
            "limit": 5
        }
    }
])

Instrumented Operations

Vector Search Detection

The instrumentor automatically detects $vectorSearch stages in aggregation pipelines and captures:

• Vector index name
• Query vector dimensions
• numCandidates
• limit

License

Apache License 2.0