[GH-ISSUE #23] Knowledge Stack & Embeddings/Vector Search Documentation #14

Closed
opened 2026-02-15 17:04:17 -05:00 by yindo · 9 comments
Owner

Originally created by @tristinbaileel on GitHub (Jan 28, 2025).
Original GitHub issue: https://github.com/cloudstack-llc/msty-docs/issues/23

As an active Msty user with experience in vector databases, embeddings, and RAG implementations, I'd like to contribute detailed documentation about Knowledge Stacks. I've noticed users asking questions about knowledge base functionality (ref: #18), and I believe comprehensive documentation could help users better utilize these features.

Proposed Documentation Topics:

  1. Knowledge Stack Fundamentals

    • What are embeddings and why they matter
    • Benefits over traditional search methods
    • Best practices for organizing knowledge bases
  2. Practical Guides

    • Setting up effective knowledge stacks
    • Choosing between local vs remote embeddings
    • Optimizing knowledge base performance
    • Integration guides (e.g., Obsidian workflow, Jina ai reranking key)
  3. Advanced Usage

    • Understanding chunk sizes and their impact
    • Tips for better search results
    • Real-world use cases and examples

I can adapt the technical depth and scope based on your preferences. My goal is to help users understand and effectively use Msty's knowledge features while maintaining any implementation details you prefer to keep internal.

Questions:

  1. Would this documentation be valuable to users?
  2. What aspects of Knowledge Stacks should be prioritized?
  3. Are there specific areas you'd like covered or avoided?
  4. What's your preferred process for documentation contributions?

I'm happy to start with a small section and iterate based on feedback.

Thank you for considering this contribution!

Originally created by @tristinbaileel on GitHub (Jan 28, 2025). Original GitHub issue: https://github.com/cloudstack-llc/msty-docs/issues/23 As an active Msty user with experience in vector databases, embeddings, and RAG implementations, I'd like to contribute detailed documentation about Knowledge Stacks. I've noticed users asking questions about knowledge base functionality (ref: #18), and I believe comprehensive documentation could help users better utilize these features. Proposed Documentation Topics: 1. Knowledge Stack Fundamentals - What are embeddings and why they matter - Benefits over traditional search methods - Best practices for organizing knowledge bases 2. Practical Guides - Setting up effective knowledge stacks - Choosing between local vs remote embeddings - Optimizing knowledge base performance - Integration guides (e.g., Obsidian workflow, Jina ai reranking key) 3. Advanced Usage - Understanding chunk sizes and their impact - Tips for better search results - Real-world use cases and examples I can adapt the technical depth and scope based on your preferences. My goal is to help users understand and effectively use Msty's knowledge features while maintaining any implementation details you prefer to keep internal. Questions: 1. Would this documentation be valuable to users? 2. What aspects of Knowledge Stacks should be prioritized? 3. Are there specific areas you'd like covered or avoided? 4. What's your preferred process for documentation contributions? I'm happy to start with a small section and iterate based on feedback. Thank you for considering this contribution!
yindo closed this issue 2026-02-15 17:04:17 -05:00
Author
Owner

@ashokgelal commented on GitHub (Jan 28, 2025):

Would this documentation be valuable to users?

I absolutely think so!

What aspects of Knowledge Stacks should be prioritized?
Are there specific areas you'd like covered or avoided?

As discussed on Discord, there is some gap in users' understanding of RAG/KS so if we could help it to educate users, despite of them being Msty users or not, that'd be great. I cannot think of anything that needs to be avoided but maybe to not get into too much technical details about vectors? And not overwhelm with telling all kinds of vector storages out there.

What's your preferred process for documentation contributions?

Preferred would be just forking this repo and adding the docs. But if running that locally is too much, I am fine with you sending just a markdown file as well (with any screenshots, if any).

Thanks for taking this initiative and looking forward to what you come up with!

@ashokgelal commented on GitHub (Jan 28, 2025): > Would this documentation be valuable to users? I absolutely think so! > What aspects of Knowledge Stacks should be prioritized? > Are there specific areas you'd like covered or avoided? As discussed on Discord, there is some gap in users' understanding of RAG/KS so if we could help it to educate users, despite of them being Msty users or not, that'd be great. I cannot think of anything that needs to be avoided but maybe to not get into too much technical details about vectors? And not overwhelm with telling all kinds of vector storages out there. > What's your preferred process for documentation contributions? Preferred would be just forking this repo and adding the docs. But if running that locally is too much, I am fine with you sending just a markdown file as well (with any screenshots, if any). Thanks for taking this initiative and looking forward to what you come up with!
Author
Owner

@tristinbaileel commented on GitHub (Jan 28, 2025):

Would this documentation be valuable to users?

I absolutely think so!

What aspects of Knowledge Stacks should be prioritized?
Are there specific areas you'd like covered or avoided?

As discussed on Discord, there is some gap in users' understanding of RAG/KS so if we could help it to educate users, despite of them being Msty users or not, that'd be great. I cannot think of anything that needs to be avoided but maybe to not get into too much technical details about vectors? And not overwhelm with telling all kinds of vector storages out there.

What's your preferred process for documentation contributions?

Preferred would be just forking this repo and adding the docs. But if running that locally is too much, I am fine with you sending just a markdown file as well (with any screenshots, if any).

Thanks for taking this initiative and looking forward to what you come up with!

Thanks for the guidance! I'll probably have something ready to share tomorrow

@tristinbaileel commented on GitHub (Jan 28, 2025): > > Would this documentation be valuable to users? > > I absolutely think so! > > > What aspects of Knowledge Stacks should be prioritized? > > Are there specific areas you'd like covered or avoided? > > As discussed on Discord, there is some gap in users' understanding of RAG/KS so if we could help it to educate users, despite of them being Msty users or not, that'd be great. I cannot think of anything that needs to be avoided but maybe to not get into too much technical details about vectors? And not overwhelm with telling all kinds of vector storages out there. > > > What's your preferred process for documentation contributions? > > Preferred would be just forking this repo and adding the docs. But if running that locally is too much, I am fine with you sending just a markdown file as well (with any screenshots, if any). > > Thanks for taking this initiative and looking forward to what you come up with! Thanks for the guidance! I'll probably have something ready to share tomorrow
Author
Owner

@tristinbaileel commented on GitHub (Jan 31, 2025):

Hello! quick question .msty ignore also respects obsidian files when connecting a vault correct?
apologies for the delay, school, work, and a newborn tends to cause those. Also for basic communicating do you prefer this or discord server?
This is how I have the structure so far in the features tab:

Image

I plan on also adding to the "how to" as well relating to obsidian integration/jina ai reranker. Each getting their own page with a slight intro of what they do etc, But i didn't know how much you wanted 3rd party services mentioned in the docs. Either way Ill respect your guidance on this, Thank you!

@tristinbaileel commented on GitHub (Jan 31, 2025): Hello! quick question .msty ignore also respects obsidian files when connecting a vault correct? apologies for the delay, school, work, and a newborn tends to cause those. Also for basic communicating do you prefer this or discord server? This is how I have the structure so far in the features tab: ![Image](https://github.com/user-attachments/assets/28d6449b-e74d-4f13-8028-90a20da10d70) I plan on also adding to the "how to" as well relating to obsidian integration/jina ai reranker. Each getting their own page with a slight intro of what they do etc, But i didn't know how much you wanted 3rd party services mentioned in the docs. Either way Ill respect your guidance on this, Thank you!
Author
Owner

@ashokgelal commented on GitHub (Jan 31, 2025):

Look good!

Not sure if I understand your question about .mstyignore but it is very similar to .gitignore (follows the exact same glob pattern). It doesn't care what kind of data source it is and ignores the files/folders that match the pattern.

I plan on also adding to the "how to" as well relating to obsidian integration/jina ai reranker. Each getting their own page with a slight intro of what they do etc, But i didn't know how much you wanted 3rd party services mentioned in the docs.

That's fine as long as it is in the context of RAG/KS and to an extent Msty.

@ashokgelal commented on GitHub (Jan 31, 2025): Look good! Not sure if I understand your question about .mstyignore but it is very similar to .gitignore (follows the exact same glob pattern). It doesn't care what kind of data source it is and ignores the files/folders that match the pattern. > I plan on also adding to the "how to" as well relating to obsidian integration/jina ai reranker. Each getting their own page with a slight intro of what they do etc, But i didn't know how much you wanted 3rd party services mentioned in the docs. That's fine as long as it is in the context of RAG/KS and to an extent Msty.
Author
Owner

@tristinbaileel commented on GitHub (Feb 1, 2025):

I tried to keep everything user-friendly while following the existing docs style.

Here's what I've created:

  1. Knowledge Stack Basics (1.basics.md)

    • Introduces the feature and what you can add (files, Obsidian vaults, YouTube content)
    • Simple getting started steps
    • Best practices and quick tips
  2. Understanding Embeddings (2.embeddings.md)

    • Explains embeddings with everyday examples (like colors to numbers)
    • Compares local vs remote options
    • Shows why it's better than regular search
  3. RAG Explained (3.rag-explained.md)

    • Breaks down RAG in simple terms
    • Shows how it works with Knowledge Stacks
    • Highlights privacy and accuracy benefits
  4. Advanced Settings (4.advanced-features.md)

    • Covers all the customization options
    • Explains chunk settings and Jina AI reranking
    • Note about screenshots: I took one of the settings interface, but it didn't look great when rendered. You might want to create a proper screenshot for advanced-chunk-settings.webp

I've used the existing components (::alert, ::tabs) and made sure everything links together nicely.

Would love your feedback on:

  • Content accuracy
  • Documentation structure
  • Any missing topics
  • Screenshot requirements

Thanks for reviewing! Happy to make any adjustments needed.

1.basics.md
2.embeddings.md
3.rag-explained.md
4.advanced-features.md

@tristinbaileel commented on GitHub (Feb 1, 2025): I tried to keep everything user-friendly while following the existing docs style. Here's what I've created: 1. **Knowledge Stack Basics** (`1.basics.md`) - Introduces the feature and what you can add (files, Obsidian vaults, YouTube content) - Simple getting started steps - Best practices and quick tips 2. **Understanding Embeddings** (`2.embeddings.md`) - Explains embeddings with everyday examples (like colors to numbers) - Compares local vs remote options - Shows why it's better than regular search 3. **RAG Explained** (`3.rag-explained.md`) - Breaks down RAG in simple terms - Shows how it works with Knowledge Stacks - Highlights privacy and accuracy benefits 4. **Advanced Settings** (`4.advanced-features.md`) - Covers all the customization options - Explains chunk settings and Jina AI reranking - Note about screenshots: I took one of the settings interface, but it didn't look great when rendered. You might want to create a proper screenshot for `advanced-chunk-settings.webp` I've used the existing components (::alert, ::tabs) and made sure everything links together nicely. Would love your feedback on: - Content accuracy - Documentation structure - Any missing topics - Screenshot requirements Thanks for reviewing! Happy to make any adjustments needed. [1.basics.md](https://github.com/user-attachments/files/18630409/1.basics.md) [2.embeddings.md](https://github.com/user-attachments/files/18630406/2.embeddings.md) [3.rag-explained.md](https://github.com/user-attachments/files/18630407/3.rag-explained.md) [4.advanced-features.md](https://github.com/user-attachments/files/18630408/4.advanced-features.md)
Author
Owner

@tristinbaileel commented on GitHub (Feb 3, 2025):

Quick question about interface screenshots for the how-to guide:

I'd like to add clear visual guides for creating Knowledge Stacks, but I want to ensure the screenshots match Msty's documentation standards. My monitor is 1440p, and my test screenshots didn't render well in the docs.

Would you be able to advise on:

  1. Preferred screenshot specifications (resolution, format, dimensions)
  2. Whether you have Figma/design files that I could reference
  3. If you'd prefer screenshots to be handled by your design team

I'm thinking of showing:

  • New Knowledge Stack creation interface
  • Import options for different content types (Files, YouTube, Obsidian)
  • Settings configuration panel

Happy to adjust my approach based on your preferences!

@tristinbaileel commented on GitHub (Feb 3, 2025): Quick question about interface screenshots for the how-to guide: I'd like to add clear visual guides for creating Knowledge Stacks, but I want to ensure the screenshots match Msty's documentation standards. My monitor is 1440p, and my test screenshots didn't render well in the docs. Would you be able to advise on: 1. Preferred screenshot specifications (resolution, format, dimensions) 2. Whether you have Figma/design files that I could reference 3. If you'd prefer screenshots to be handled by your design team I'm thinking of showing: - New Knowledge Stack creation interface - Import options for different content types (Files, YouTube, Obsidian) - Settings configuration panel Happy to adjust my approach based on your preferences!
Author
Owner

@ashokgelal commented on GitHub (Feb 3, 2025):

I don't have preferred sizes other than the sizes of the images that we are already using. You can look at them for the reference. However, if you only have 1440p monitor, the screenshots might not come out good. Let's do this - you add row resolution screenshots and before publish we'll replace them with high res (making them as close to your original source as possible.

@ashokgelal commented on GitHub (Feb 3, 2025): I don't have preferred sizes other than the sizes of the images that we are already using. You can look at them for the reference. However, if you only have 1440p monitor, the screenshots might not come out good. Let's do this - you add row resolution screenshots and before publish we'll replace them with high res (making them as close to your original source as possible.
Author
Owner

@tristinbaileel commented on GitHub (Feb 4, 2025):

Added/Updated the docs some more!

What's New

  • Added a detailed how-to guide: create-knowledge-stacks.md
  • Created 10 interface screenshots showing every part of the Knowledge Stack feature
  • Rewrote the four main Knowledge Stack docs to be clearer and more accurate

Files Added

New Guide & Screenshots

  • how-to guides/create-knowledge-stacks.md
  • added screenshots for public/how-to/create-knowledge-stack/

Updated Docs

  1. features/knowledge-stack/1.basics.md
  2. features/knowledge-stack/2.embeddings.md
  3. features/knowledge-stack/3.rag-explained.md
  4. features/knowledge-stack/4.advanced-features.md

Note: These are complete rewrites of the feature docs, so you can discard the original versions.

Main Improvements

  • Added all missing features and settings
  • Included actual default values
  • Added clear examples and use cases
  • Made sure everything matches the current interface
  • Connected all docs with proper cross-references

Updated Files:
1.basics.md
2.embeddings.md
3.rag-explained.md
4.advanced-features.md

New File:
create-knowledge-stacks.md

Needed Images for files:

custom-notes-import.webp:
Image

file-import.webp:
Image

folder-import.webp:
Image

new-stack-button.webp:
Image

obsidian-import.webp:
Image

search-settings-interface.webp:
Image

youtube-import.webp:
Image

choose-knowledge-stack-in-chat-interface.webp:
Image

chunk-settings-interface.webp:
Image

compose-save-as-draft-buttons.webp:
Image

@tristinbaileel commented on GitHub (Feb 4, 2025): Added/Updated the docs some more! ### What's New - Added a detailed how-to guide: `create-knowledge-stacks.md` - Created 10 interface screenshots showing every part of the Knowledge Stack feature - Rewrote the four main Knowledge Stack docs to be clearer and more accurate ### Files Added **New Guide & Screenshots** - `how-to guides/create-knowledge-stacks.md` - added screenshots for `public/how-to/create-knowledge-stack/` **Updated Docs** 1. `features/knowledge-stack/1.basics.md` 2. `features/knowledge-stack/2.embeddings.md` 3. `features/knowledge-stack/3.rag-explained.md` 4. `features/knowledge-stack/4.advanced-features.md` Note: These are complete rewrites of the feature docs, so you can discard the original versions. ### Main Improvements - Added all missing features and settings - Included actual default values - Added clear examples and use cases - Made sure everything matches the current interface - Connected all docs with proper cross-references Updated Files: [1.basics.md](https://github.com/user-attachments/files/18665302/1.basics.md) [2.embeddings.md](https://github.com/user-attachments/files/18665301/2.embeddings.md) [3.rag-explained.md](https://github.com/user-attachments/files/18665300/3.rag-explained.md) [4.advanced-features.md](https://github.com/user-attachments/files/18665299/4.advanced-features.md) New File: [create-knowledge-stacks.md](https://github.com/user-attachments/files/18665304/create-knowledge-stacks.md) Needed Images for files: custom-notes-import.webp: ![Image](https://github.com/user-attachments/assets/1d105fde-6b19-4f50-bc11-8b9fb23ad9b2) file-import.webp: ![Image](https://github.com/user-attachments/assets/fff223ef-37bc-4639-8ff6-2de23a72f17a) folder-import.webp: ![Image](https://github.com/user-attachments/assets/70a3ff29-f63b-4e83-a9bc-a0f7ce485b50) new-stack-button.webp: ![Image](https://github.com/user-attachments/assets/56f11df1-3ee3-4de9-9f99-b8bcd0e99553) obsidian-import.webp: ![Image](https://github.com/user-attachments/assets/d11ad6f0-56fc-4e1f-b08f-465be9741e60) search-settings-interface.webp: ![Image](https://github.com/user-attachments/assets/45a72c18-1758-42b4-8c7c-9abca2e6f315) youtube-import.webp: ![Image](https://github.com/user-attachments/assets/8078ba8f-d74e-4fe7-8ff2-76258af4b0a6) choose-knowledge-stack-in-chat-interface.webp: ![Image](https://github.com/user-attachments/assets/d36aa62b-51be-4e43-a648-3854a1cc574b) chunk-settings-interface.webp: ![Image](https://github.com/user-attachments/assets/7237af62-93fe-40bd-8336-87bac468502d) compose-save-as-draft-buttons.webp: ![Image](https://github.com/user-attachments/assets/720bcacb-f8a8-4e99-a680-86b851200ea6)
Author
Owner

@ashokgelal commented on GitHub (Feb 7, 2025):

It has now been merged! Thank you so much for your contribution @tristinbaileel!

https://docs.msty.app/features/knowledge-stack/basics

@ashokgelal commented on GitHub (Feb 7, 2025): It has now been merged! Thank you so much for your contribution @tristinbaileel! https://docs.msty.app/features/knowledge-stack/basics
yindo changed title from Knowledge Stack & Embeddings/Vector Search Documentation to [GH-ISSUE #23] Knowledge Stack & Embeddings/Vector Search Documentation 2026-06-05 16:26:59 -04:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: cloudstack-llc/msty-docs#14