Proposed: MindManager + Obsidian + AI using MCP servers

krsto e. shared this idea 47 days ago
Voting Open

77481055ee93de170941516d81bcf623cf1b081238ac734267a2d9e02148a4a3



I want to merge MindManager + Obsidian + AI tools into one workflow. I have a partial solution working and proposing how to complete it.

What works now:

- I have a VBA macro that exports .mmap → .md files

- AI can read the .md files

- I can view/edit .md in Obsidian/VS Code

What's missing:

- AI can't write changes BACK into the .mmap file

- Example: "AI, reorganize my study topics" → needs to edit the .mmap directly

Proposed solution - extend existing MCP server:

There's a Python library "mindm" by robertZaufall that already has all the functions we need:

- set_text_to_topic() - edit topic text

- add_subtopic() - add new topics

- add_note() - add notes

- create_relationship() - link topics

- add_tag() - tag topics

See: https://github.com/robertZaufall/mindm

He also made an MCP server (https://github.com/robertZaufall/mindm-mcp) but it currently only exposes READ functions.

The fix: Wrap his existing Python WRITE functions into MCP tools. MCP tools are just Python functions exposed to AI, so the extension should be straightforward if someone has time.

Result: Complete workflow - visual thinking in MM, text notes in Obsidian, AI handles reorganization/generation.

[Screenshot of workflow graphic]

Has anyone tried this? Any feedback before I start extending the MCP server?

Happy to share my export macro if useful.

Replies (6)

photo
1

keeping an intermediate file as backup is good.

so...

8b24d5b1194669e15c4fe632db89957f31d7834cf99f9470aeee465deb6df67e

8ef53bd67bb4c13bbd813dbf5a027debe3770217f5cd6992824a61c776a6714a


2e3bf3150eccb5a015d7d2a5cc139200d811d028f4607b712cf7fba165f9fa23


ef31995a813dfba74abb11e6b3cd6a9c871ee5d2f87a5e1dca79f9709799b829


i propose that later, there is cardiology.mmap, and all the other files are in a folder cardiology.mmap.assets;

this is alike in typora or tagspaces to keep the sidecar files.

photo
1

so now let's see the author of the MCP server have time or who else would have time for that.

thanks

photo
1

cce299de524ffdd0588a77f0c4cdcabd61de8d425c245becce14e71a6fbfe3af

photo
1

ps i dont know why my ideas already end up in "questions", i think when create new topic, one need to choose it manually.

photo
1

251113T

Hi everyone,

I wanted to share my experience setting up an integration between MindManager 2020 and Cursor IDE using MCP servers in a Hyper-V Windows 10 VM.

Current Working Setup:

MindManager 2020 + Cursor IDE

mindm and mindm-mcp servers installed

LLMs in Cursor IDE can interact with MindManager through the MCP servers

Current State:

The mindm-mcp server (a wrapper around the mindm Python library) currently provides 9 tools, primarily for reading mindmap data. I've tested this functionality and it works well.

What's Missing:

The other half of the mindm Python library needs to be wrapped into the mindm-mcp server to enable write functions.

My Progress:

Within 5 minutes using Cursor's free "auto" LLM, I successfully created a "create_topic" MCP tool. This demonstrates how straightforward the development process can be.

Development Plans:

I contacted the original author about contributing, but he only accepts PRs that work on both Windows and macOS. Since I don't have access to macOS, I'll be developing my own complementary MCP server named "mindm_20+" (or similar) to avoid conflicts with the original project. This approach:

Allows the original author to potentially add write functions later with higher code quality

Provides temporary write functionality for users who need it now

Specifically supports MindManager 2020 (and later), the last version that works in air-gapped environments (newer versions require internet connection every 30 days)

Use Case:

This integration will enable seamless conversion between .mindmap files and .md files for Obsidian, with LLMs able to make modifications directly to .mindmap files.

I'll be working on this as a hobby project. Since it's essentially wrapping existing work from the mindm library, development should be relatively straightforward.

Anyone interested in testing or contributing is welcome to reach out!

Leave a Comment
 
Attach a file
You can't vote. Please authorize!