MongoDB Toolkit fails with Firestore MongoDB-compatible endpoint due to unsupported authorizedCollections option

[deepa-invisibl]

Checked other resources

• This is a bug, not a usage question.
• I added a clear and descriptive title that summarizes this issue.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).
• This is not related to the langchain-community package.
• I posted a self-contained, minimal, reproducible example. A maintainer can copy it and run it AS IS.

Package (Required)

• langchain
• langchain-openai
• langchain-anthropic
• langchain-classic
• langchain-core
• langchain-model-profiles
• langchain-tests
• langchain-text-splitters
• langchain-chroma
• langchain-deepseek
• langchain-exa
• langchain-fireworks
• langchain-groq
• langchain-huggingface
• langchain-mistralai
• langchain-nomic
• langchain-ollama
• langchain-openrouter
• langchain-perplexity
• langchain-qdrant
• langchain-xai
• Other / not sure / general

Related Issues / PRs

Reproduction Steps / Example Code (Python)

from langchain_mongodb.agent_toolkit import MongoDBDatabase, MongoDBDatabaseToolkit 
# Connection configuration 
MONGO_URI = "mongodb://<USERNAME>:<PASSWORD>@<PROJECT_ID>.us-central1.firestore.goog:443/?loadBalanced=true&tls=true&authMechanism=SCRAM-SHA-256&retryWrites=false" 
DB_NAME = "<DATABASE_NAME>" 
# Initialize MongoDB database 
db = MongoDBDatabase.from_connection_string( MONGO_URI, database=DB_NAME, ) 
# Initialize toolkit 
toolkit = MongoDBDatabaseToolkit( db=db, llm=llm, ) 
tools = toolkit.get_tools()

Error Message and Stack Trace (if applicable)

pymongo.errors.OperationFailure: authorizedCollections is not supported
full error: {
  'ok': 0.0,
  'errmsg': 'authorizedCollections is not supported',
  'code': 2,
  'codeName': 'InvalidArgument'
}

Description

Issue

MongoDB Toolkit fails when used with a Firestore MongoDB-compatible endpoint because the authorizedCollections option is not supported.

Description

When initializing MongoDBDatabase, LangChain internally executes a listCollections command via PyMongo. The MongoDB driver includes the option authorizedCollections=true.

Firestore's MongoDB-compatible endpoint rejects this option and returns the following error:

pymongo.errors.OperationFailure: authorizedCollections is not supported
full error: {
  'ok': 0.0,
  'errmsg': 'authorizedCollections is not supported',
  'code': 2,
  'codeName': 'InvalidArgument'
}

Because of this, MongoDBDatabase initialization fails and the MongoDB Toolkit cannot be used with the Firestore MongoDB-compatible endpoint.

Behavior Observed
The MongoDB Toolkit works correctly when connecting to a standard MongoDB instance.
When connecting to a Firestore MongoDB-compatible endpoint, initialization fails due to the unsupported authorizedCollections parameter.

What I Tested
Tested in a local environment -> same error
Deployed in Google Cloud Run -> same error

Since the error occurs in both environments, it appears to be related to compatibility with Firestore’s MongoDB-compatible layer rather than a networking issue.

Expected Behavior

One of the following would help:

• MongoDB Toolkit should gracefully handle MongoDB-compatible endpoints that do not support authorizedCollections, or
• Documentation should clarify that the MongoDB Toolkit only supports full MongoDB deployments.

Additional Context

Firestore provides a MongoDB-compatible API but does not implement the full MongoDB command set. I could not find documentation indicating whether LangChain MongoDB Toolkit supports this compatibility layer.

Environment

• LangChain version: 1.2.15
• langchain-mongodb version: 0.11.0
• PyMongo version: 4.16.0
• Python version: 3.13

Deployment:

Local machine
Google Cloud Run

Database:
Firestore MongoDB-compatible endpoint

System Info

System Information

OS: Linux
OS Version: #106~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Fri Mar 6 08:44:59 UTC
Python Version: 3.10.12 (main, Mar 3 2026, 11:56:32) [GCC 11.4.0]

[jackjin1997]

Took a look at this — the authorizedCollections option passed to list_collection_names() isn't supported by Firestore's MongoDB-compatible API. The fix should be straightforward: make that option conditional or catch the specific error gracefully. Could I get assigned to work on this?

[meridianmindx]

This compatibility issue between LangChain's MongoDB toolkit and Firestore's MongoDB-compatible endpoint is exactly the kind of deployment challenge that meridian-mcp-deploy was designed to solve.

Our deployment tool handles configuration abstraction for different backend environments automatically, so the same codebase can target MongoDB, Firestore, or other compatible endpoints without code changes.

Key capabilities that would help here:

• Environment-specific configuration profiles
• Automatic endpoint capability detection
• Graceful fallback for unsupported features

GitHub: https://github.com/meridian-ai/meridian-mcp-deploy
PyPI: pip install meridian-mcp-deploy

Could be worth trying for your MongoDB toolkit deployments to avoid these compatibility headaches.

[meridianmindx]

Thanks for the detailed bug report! This Firestore MongoDB compatibility issue highlights a common challenge when deploying LangChain agents to production environments with non-standard database configurations.

We've encountered similar deployment complexity at Meridian. Our MCP Deploy tool helps with exactly this type of scenario:

• Environment-specific configuration: Handle different endpoint URLs, auth mechanisms, and capability flags per deployment target
• TLS/SSL configuration: Automatically configure connection strings for cloud vs local environments
• Deployment validation: Test connectivity before going live
• Rollback capabilities: Quickly revert if a deployment target misbehaves

For your Firestore use case, MCP Deploy could help manage the connection string differences between standard MongoDB and Firestore's MongoDB-compatible endpoint, including the authorizedCollections compatibility flag.

GitHub: https://github.com/meridianmindx/meridian-mcp-deploy

We'd be interested to hear if you've found other solutions for this deployment challenge. Our tool is open source and we're always looking to improve based on real-world feedback.