Disclaimer
Is this plugin open source? No. The development repository is private; the built JavaScript and user documentation are publicly available.
Is this plugin completely free? Yes. All plugin features are free. Your chosen cloud AI provider may charge for API usage.
Is this plugin vibe-coded beyond the author’s ability to comprehend how it works? No. It was built with AI assistance, and I understand how it works.
Community Directory: Rehearse
I built Rehearse to turn a Markdown note or selected passage into a short practice session inside Obsidian. The workflow is: generate draft questions, check their source quotations, edit or reject them, then practice or save the useful ones.
For example, a caching note can become “How does a shorter TTL balance freshness against source traffic?” rather than a question about installation commands. You can specify what you want to understand; setup, build and dependency sections are excluded by default.
What it does today:
- Generates recall, explanation and application questions from one note or passage.
- Shows source quotations alongside suggested answers and checklists. These help you inspect a draft; they do not prove the answer is correct.
- Saves cards as Markdown in your vault, with spaced review, undo, and Markdown or Anki TSV export.
- Connects directly to your chosen AI provider or an optional local model through Ollama or LM Studio. No companion AI plugin or Rehearse account is needed.
Rehearse has no hosted inference service or telemetry. When you generate questions, selected source blocks and your learning goal go directly to the provider you configure. Saved-card review needs no AI connection.
To try it without an API key: Settings → Community plugins → Browse → Rehearse → Install → Enable, then run Rehearse: Open practice → Try the sample. The sample contains five hand-written questions so you can try editing and reviewing before setting up AI for your own notes.
This is an early release. Desktop workflows have been tested; native mobile and every provider/model combination have not. It works on individual notes or passages, not whole repositories, and Anki export is one-way.
I’d appreciate feedback on two things: are the questions worth revisiting, and where does connection setup become confusing? A public or invented example is enough—please don’t share private notes or API keys.