Implement MongoDB Vector Search Node in n8n.io

XMLWordPrintableJSON

    • Type: Epic
    • Resolution: Unresolved
    • Priority: Major - P3
    • None
    • Affects Version/s: None
    • Component/s: None
    • Vector Search in n8n.io
    • Node Drivers
    • None
    • Hide
      1. What would you like to communicate to the user about this feature?
      2. Would you like the user to see examples of the syntax and/or executable code and its output?
      3. Which versions of the driver/connector does this apply to?
      Show
      1. What would you like to communicate to the user about this feature? 2. Would you like the user to see examples of the syntax and/or executable code and its output? 3. Which versions of the driver/connector does this apply to?
    • To Do
    • 5
    • 5.2
    • 6
    • 100
    • 20
    • 🟡 Potential Risk
    • Hide

      Known risks or blockers:

      • Delete support for embedded documents is waiting on their AI team to review the PR, so we are pushing the calendar date out another two weeks.

      Completed over the last 2 weeks:

      • n8n merged the two open PRs we had: the metadata and pre/post filter support for vector search queries
      • we were unable to go with the original CRUD approach and we did not get much guidance from n8n for implementing the required general delete mode for all nodes (not just MongoDB), so we opened a PR with our best effort to do the latter along with the implementation of delete in our vector search node

      Focus over the next 2 weeks:

      • Ideally, get feedback from n8n, address comments and merge the delete PR, which would complete this project.
      Show
      Known risks or blockers: Delete support for embedded documents is waiting on their AI team to review the PR, so we are pushing the calendar date out another two weeks. Completed over the last 2 weeks: n8n merged the two open PRs we had: the metadata and pre/post filter support for vector search queries we were unable to go with the original CRUD approach and we did not get much guidance from n8n for implementing the required general delete mode for all nodes (not just MongoDB), so we opened a PR with our best effort to do the latter along with the implementation of delete in our vector search node Focus over the next 2 weeks: Ideally, get feedback from n8n, address comments and merge the delete PR, which would complete this project.
    • Hide

      2025-09-02 - 🟡 Potential Risk
      Known risks or blockers:

      • Delete support for embedded documents is blocked, waiting on n8n to provide input about how to implement this feature.

      Completed over the last 2 weeks:

      • Work was started to integrate n8n testing in the AI testing pipeline.
      • We added client metadata tracking to the n8n project.  This work is in progress: PR reviewed, comments addressed and waiting for n8n merge.
      • pre/post filter support for vector search has comments addressed, waiting for n8n to merge.

      Focus over the next 2 weeks:

      • We'd like to merge the open PRs we have against n8n.
      • We'd like to get unblocked by n8n for `delete` operations and start implementing this feature.

      Engineer(s): Durran Jordan
      2025-08-15: Target date set to 2025-09-12

      Known risks or blockers:

      • We don't see a way to add a delete operation to the vector search Node (NODE-6915); we will need to reach out to the n8n team for assistance with this issue.
      • n8n said they would review any of our PRs next week, but they have said this before, so we are conservatively pushing the calendar date for completion back another couple of weeks

      Completed over the last 2 weeks:

      • n8n merged the search index node refactor PR, unblocking the already prepared PR for pre/post search index support.
      • n8n have given us the green light to also submit the next and final functional feature for this project (delete operations for embedded documents) concurrently.

      Focus over the next 2 weeks:

      • At minimum, we will rebase, clean up, and open the pre/post PR. Ideally, we'll find a way to implement NODE-6915 and open that for review as well.

      Engineer(s): Bailey Pearson, Durran Jordan
      2025-08-01: Target date unchanged

      Known risks or blockers:

      • Calendar delays due to external review coordination with n8n: they have only reviewed one of the three PRs we have put up.

      Completed over the last 2 weeks:

      • N/A - work was paused waiting for n8n review

      Focus over the next 2 weeks:

      • Addressing feedback on the pull request that received feedback. Addressing other feedback if it is provided.

      Engineer(s): Bailey Pearson, Durran Jordan
      2025-07-18: Target date set to 2025-08-29

      Known risks or blockers:

      • Calendar delays are due to external review coordination with n8n: they still have not reviewed any of the PRs we have put up; they said (again) that they "might" be able to review them next week (week of 7/21, previously they told us the same thing about 7/7).

      Completed over the last 2 weeks:

      • N/A - work paused waiting for n8n review

      Focus over the next 2 weeks:

      • Address any feedback from the n8n team, assuming we get it, then continue with the second half of the epic

      Engineer(s): Bailey Pearson, Durran Jordan
      2025-07-03: Target date set to 2025-07-25

      Known risks or blockers:

      • Calendar delays are due to external review coordination with n8n: they still have not reviewed any of the PRs we have put up, they say they "might" be able to review them next week (week of 7/7).

      Completed over the last 2 weeks:

      • N/A - work paused waiting for n8n review

      Focus over the next 2 weeks:

      • Address any feedback from the n8n team, assuming we get it, then continue with the second half of the epic

      Engineer(s): Bailey Pearson, Durran Jordan
      2025-06-20: Target date unchanged

      Known risks or blockers:

      • Potential calendar delays due to external review needed (n8n): they have not reviewed any of the PRs we have put up, and attempts to get their attention via Prakul have not been successful

      Completed over the last 2 weeks:

      • Three pull requests up for best practices audit fixups, adding a new node for search index management, and support for pre/post search queries

      Focus over the next 2 weeks:

      • We are pausing work until the existing PRs get feedback from the n8n team

      Engineer(s): Bailey Pearson, Durran Jordan
      2025-06-06: Target date set to 2025-07-11

      Known risks or blockers:

      • Potential calendar delays due to external review needed (n8n)

      Completed over the last 2 weeks:

      • Project started and getting set up. One PR for the audit of the vector search node needs review.

      Focus over the next 2 weeks:

      • Search index management and updates to the CRUD node.
      Show
      2025-09-02 - 🟡 Potential Risk Known risks or blockers: Delete support for embedded documents is blocked, waiting on n8n to provide input about how to implement this feature. Completed over the last 2 weeks: Work was started to integrate n8n testing in the AI testing pipeline. We added client metadata tracking to the n8n project.  This work is in progress: PR reviewed, comments addressed and waiting for n8n merge. pre/post filter support for vector search has comments addressed, waiting for n8n to merge. Focus over the next 2 weeks: We'd like to merge the open PRs we have against n8n. We'd like to get unblocked by n8n for `delete` operations and start implementing this feature. Engineer(s): Durran Jordan 2025-08-15: Target date set to 2025-09-12 Known risks or blockers: We don't see a way to add a delete operation to the vector search Node ( NODE-6915 ); we will need to reach out to the n8n team for assistance with this issue. n8n said they would review any of our PRs next week, but they have said this before, so we are conservatively pushing the calendar date for completion back another couple of weeks Completed over the last 2 weeks: n8n merged the search index node refactor PR, unblocking the already prepared PR for pre/post search index support. n8n have given us the green light to also submit the next and final functional feature for this project (delete operations for embedded documents) concurrently. Focus over the next 2 weeks: At minimum, we will rebase, clean up, and open the pre/post PR. Ideally, we'll find a way to implement NODE-6915 and open that for review as well. Engineer(s): Bailey Pearson, Durran Jordan 2025-08-01: Target date unchanged Known risks or blockers: Calendar delays due to external review coordination with n8n: they have only reviewed one of the three PRs we have put up. Completed over the last 2 weeks: N/A - work was paused waiting for n8n review Focus over the next 2 weeks: Addressing feedback on the pull request that received feedback. Addressing other feedback if it is provided. Engineer(s): Bailey Pearson, Durran Jordan 2025-07-18: Target date set to 2025-08-29 Known risks or blockers: Calendar delays are due to external review coordination with n8n: they still have not reviewed any of the PRs we have put up; they said (again) that they "might" be able to review them next week (week of 7/21, previously they told us the same thing about 7/7). Completed over the last 2 weeks: N/A - work paused waiting for n8n review Focus over the next 2 weeks: Address any feedback from the n8n team, assuming we get it, then continue with the second half of the epic Engineer(s): Bailey Pearson, Durran Jordan 2025-07-03: Target date set to 2025-07-25 Known risks or blockers: Calendar delays are due to external review coordination with n8n: they still have not reviewed any of the PRs we have put up, they say they "might" be able to review them next week (week of 7/7). Completed over the last 2 weeks: N/A - work paused waiting for n8n review Focus over the next 2 weeks: Address any feedback from the n8n team, assuming we get it, then continue with the second half of the epic Engineer(s): Bailey Pearson, Durran Jordan 2025-06-20: Target date unchanged Known risks or blockers: Potential calendar delays due to external review needed (n8n): they have not reviewed any of the PRs we have put up, and attempts to get their attention via Prakul have not been successful Completed over the last 2 weeks: Three pull requests up for best practices audit fixups, adding a new node for search index management, and support for pre/post search queries Focus over the next 2 weeks: We are pausing work until the existing PRs get feedback from the n8n team Engineer(s): Bailey Pearson, Durran Jordan 2025-06-06: Target date set to 2025-07-11 Known risks or blockers: Potential calendar delays due to external review needed (n8n) Completed over the last 2 weeks: Project started and getting set up. One PR for the audit of the vector search node needs review. Focus over the next 2 weeks: Search index management and updates to the CRUD node.
    • 5

      Develop a new node (connector) within the n8n framework (a third-party open-source app dev framework) to enable integration with MongoDB for vector search functionality. This feature will allow n8n users to leverage MongoDB as a vector store, facilitating advanced similarity searches for vectorized data. 

      Use Case

      As a developer using n8n
      I want to be able to use MongoDB Vector Search
      So that I can build AI applications like natural language chatbots, semantic search engines, etc.

      Customer notes which talk about some N8n requests - https://docs.google.com/document/d/14vrxlRdkuDTjbtXKCqudxHkZeQOvTFy4OXJQ6ounDz0/edit?pli=1&tab=t.50y9ug4ihhhn#heading=h.tpqd0cdo1vlu

      Dependencies

      None

      Risks/Unknowns

      None

      Acceptance Criteria

      Implementation Requirements

      Use MongoDB vector store LangChain integration as a reference.

      Connection Setup:

      • Allow users to configure MongoDB connection settings, including hostname, port, database, username, and password.

      Index Management:

      • Create, manage, and delete vector indexes. Allow specifying vector dimensions and distance metrics (e.g., cosine similarity, Euclidean).

      CRUD Operations:

      • Insert vectorized data into MongoDB collections with metadata. Update and delete entries.  

      Query Support:

      • Perform similarity searches with user-defined query vectors. Support filters based on metadata (e.g., filtering by category or tags).

      Testing Requirements

      • Add standard unit tests and add them to MongoDB Evergreen testing suite
      • Integrate with the MongoDB AI/ML pipeline

      Documentation Requirements

      • Docs ticket dependency for adding to "AI Integration" section  in vector search doc

       

       

      Additional links:

      https://docs.n8n.io/integrations/creating-nodes/overview/

      https://docs.n8n.io/help-community/contributing/

      https://github.com/n8n-io/n8n/tree/master/packages/%40n8n/nodes-langchain/nodes/vector_store/VectorStorePinecone

            Assignee:
            Bailey Pearson
            Reporter:
            Prakul Agarwal
            Daria Pardue Daria Pardue
            None
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

              Created:
              Updated:
              16 weeks, 2 days
              None
              None
              None