bart6114/loofah

Free, open-source Markdown knowledge vault for macOS. Record and transcribe meetings locally, keep notes as files you can carry across tools, and give agents read-only MCP access. No accounts, subscriptions, telemetry, or cloud backend.

6

stars

8,615

commits

Rust

primary language

Sep 9, 2026

updated

loofah.io
knowledge-management
local-first
macos
markdown
mcp
meeting-notes
notetaking
personal-knowledge-management
plain-text
privacy
speech-to-text
stt
tauri
transcription
Browse cluster: Whisper Speech Recognition & Deployment

README

Loofah app icon

Loofah

Download for macOS (Apple Silicon)

I wanted a meeting notetaker that did a few things well: transcribe locally, keep my notes as ordinary Markdown files, and stay out of the way. I couldn't find one without accounts, subscriptions, or a cloud service in the middle, so I built Loofah on anarlog.

Loofah showing a meeting brief, notes, and recent sessions in a local vault

There is no backend, account, billing, telemetry, or premium tier. Your notes live on your computer and remain usable without this app. If you want AI-generated summaries, connect your own LLM—OpenAI, Anthropic, Gemini, OpenRouter, Ollama, LM Studio, or another OpenAI-compatible provider.

This is software I built for my own day-to-day use. It may be rough around the edges, but if you want the same kind of tool, I hope it is useful to you too.

Getting started

Download the signed and notarized Apple Silicon build using the button above. That link always points to the latest published version; previous versions are available on the releases page. You can also build it from source using the instructions below.

Open the app, start a session, and join your meeting. The app records and transcribes on your Mac, then saves the notes as Markdown in your vault.

What matters here

  • The files are yours. Each meeting is stored as Markdown that you can read, search, edit, back up, or sync with whatever you already use.
  • Transcription stays local. Your meeting audio does not need to be sent to a transcription service.
  • AI is optional and bring-your-own. Use a hosted provider or run a local model with Ollama or LM Studio.
  • No account or tracking. Install the app and use it. There is nothing to sign up for.
  • CLI and MCP support. The included loof CLI and MCP server can give your scripts and coding agents read-only access to meeting notes.

Development

This is a pnpm workspace containing a Tauri desktop app (apps/desktop/) and a Rust CLI (apps/cli/). There is no database: the files in the vault are the source of truth. The vault format lives in crates/vault-read/; the interface uses Zustand for state and TipTap for editing.

To run it locally:

pnpm install
pnpm -F @hypr/desktop tauri:dev   # run the desktop app
cargo build -p loof-cli              # build the loof CLI

See AGENTS.md for development notes, including formatting, typechecking, and code-style conventions.

License

MIT. See LICENSE for the complete license and copyright history, including the original anarlog copyright.

Contributors

(top 30 of 33)

yujonglee

4,718 commits

ComputelessComputer

2,106 commits

bart6114

332 commits

bart6114/loofah

Free, open-source Markdown knowledge vault for macOS. Record and transcribe meetings locally, keep notes as files you can carry across tools, and give agents read-only MCP access. No accounts, subscriptions, telemetry, or cloud backend.

6

stars

8,615

commits

Rust

primary language

Sep 9, 2026

updated

loofah.io
knowledge-management
local-first
macos
markdown
mcp
meeting-notes
notetaking
personal-knowledge-management
plain-text
privacy
speech-to-text
stt
tauri
transcription
Browse cluster: Whisper Speech Recognition & Deployment

README

Loofah app icon

Loofah

Download for macOS (Apple Silicon)

I wanted a meeting notetaker that did a few things well: transcribe locally, keep my notes as ordinary Markdown files, and stay out of the way. I couldn't find one without accounts, subscriptions, or a cloud service in the middle, so I built Loofah on anarlog.

Loofah showing a meeting brief, notes, and recent sessions in a local vault

There is no backend, account, billing, telemetry, or premium tier. Your notes live on your computer and remain usable without this app. If you want AI-generated summaries, connect your own LLM—OpenAI, Anthropic, Gemini, OpenRouter, Ollama, LM Studio, or another OpenAI-compatible provider.

This is software I built for my own day-to-day use. It may be rough around the edges, but if you want the same kind of tool, I hope it is useful to you too.

Getting started

Download the signed and notarized Apple Silicon build using the button above. That link always points to the latest published version; previous versions are available on the releases page. You can also build it from source using the instructions below.

Open the app, start a session, and join your meeting. The app records and transcribes on your Mac, then saves the notes as Markdown in your vault.

What matters here

  • The files are yours. Each meeting is stored as Markdown that you can read, search, edit, back up, or sync with whatever you already use.
  • Transcription stays local. Your meeting audio does not need to be sent to a transcription service.
  • AI is optional and bring-your-own. Use a hosted provider or run a local model with Ollama or LM Studio.
  • No account or tracking. Install the app and use it. There is nothing to sign up for.
  • CLI and MCP support. The included loof CLI and MCP server can give your scripts and coding agents read-only access to meeting notes.

Development

This is a pnpm workspace containing a Tauri desktop app (apps/desktop/) and a Rust CLI (apps/cli/). There is no database: the files in the vault are the source of truth. The vault format lives in crates/vault-read/; the interface uses Zustand for state and TipTap for editing.

To run it locally:

pnpm install
pnpm -F @hypr/desktop tauri:dev   # run the desktop app
cargo build -p loof-cli              # build the loof CLI

See AGENTS.md for development notes, including formatting, typechecking, and code-style conventions.

License

MIT. See LICENSE for the complete license and copyright history, including the original anarlog copyright.

Contributors

(top 30 of 33)

yujonglee

4,718 commits

ComputelessComputer

2,106 commits

bart6114

332 commits

Languages

Rust

50.1%

TypeScript

46.7%

Swift

2.0%