Obsidian_fetch
Retrieve and load notes efficiently from Obsidian vaults, enabling enhanced interactions with language models by cleaning link queries and displaying backlinks to opened files. Streamlined for local GPU setups to improve note retrieval speed and efficiency.
Author

soukouki
Quick Info
Actions
Tags
ObsidianFetch
MCP servers focused on fetching and presenting information from Obsidian vaults.
The existing MCP server has the following drawbacks:
- It supports many commands, which can cause slow prompt loading when computational resources are limited.
- When reading a note labeled "Sample Note", it is necessary to search for its path first before loading it, but the LLM may not always follow this procedure.
- Some tools include unnecessary options, leading the LLM to sometimes fail to invoke them correctly.
These issues become particularly noticeable when running an LLM on a local GPU.
To address this, we developed a new MCP server that simply retrieves and loads lists of notes.
The new server also provides the following additional features:
- When the LLM attempts to retrieve link information by searching with brackets like
[[link name]]
, the server automatically removes any characters that cannot be used in links. - In addition to loading the note contents, it also displays backlinks—notes that link to the currently opened note.
- This allows the LLM to load and understand the connections between related notes via backlinks.
Installation
gem install obsidian_fetch
Usage
obsidian_fetch /path/to/your/vault
Contributing
Bug reports and pull requests are welcome on GitHub at https://github.com/soukouki/obsidian_fetch.
License
The gem is available as open source under the terms of the MIT License.