AI Vault for Obsidian

Chat with GPT and Claude directly inside Obsidian — with RAG over your vault, conversation history, projects, and smart modes.

AI Vault turns Obsidian into a full AI workspace. Talk to multiple frontier models, give them access to your notes and canvases, organize chats into projects, and keep everything synced with your vault — without ever leaving the app.

---

✨ Features

🤖 Multi-provider AI

Switch between OpenAI and Anthropic models with a single click in the chat header — no need to dive into settings.

OpenAI models:

• GPT-5
• GPT-5 nano
• GPT-5 mini
• GPT-5 Search
• GPT-4o
• GPT-4o mini
• GPT-4o turbo

Anthropic models:

• Claude Sonnet 4.5
• Claude Opus 4.5
• Claude Haiku 4.5

📚 RAG over your vault

The plugin indexes your notes locally and feeds the most relevant context to the AI on every message.

• Full support for Markdown (.md) files
• Full support for Canvas (.canvas) files — the parser understands canvas topology through edges and reconstructs the logical flow of nodes
• Local index stored outside the vault (no sync bloat)

🗂️ Projects

Group related conversations into projects — each with its own system prompt, model, and context scope. Perfect for separating work, study, and personal threads.

💬 Conversation history

Every chat is saved automatically and stored outside the vault, so it never inflates your Obsidian Sync or clutters your file tree.

🎯 Smart modes

Switch the AI’s behavior depending on what you need:

• Code mode — focused on programming, with code-aware formatting
• Learn mode — explains concepts step by step, asks clarifying questions
• Thinking mode — deeper reasoning before answering

🌐 Web search

Use GPT-5 Search to fetch live information from the web directly in your chat.

🌍 Bilingual UI

Full interface in English and Polish.

⚡ Streaming & cost tracking

• Live token streaming — see the response as it’s generated
• Built-in token counter and cost tracker for every conversation

---

📱 Platform support

Platform	Status
Windows	✅ Supported
macOS	✅ Supported
Linux	✅ Supported

---

🚀 Installation

From Obsidian Community Plugins (recommended once approved)

1. Open Settings → Community plugins
2. Make sure Restricted mode is off
3. Click Browse and search for AI Vault
4. Click Install, then Enable

Manual installation

1. Download the latest release from the Releases page

2. Extract main.js, manifest.json, and styles.css into:

   <your-vault>/.obsidian/plugins/ai-vault/
   

3. Reload Obsidian

4. Enable AI Vault in Settings → Community plugins

---

🔑 Setup

After enabling the plugin:

1. Open Settings → AI Vault
2. Paste your OpenAI API key and/or Anthropic API key
3. Choose your default model
4. Open the chat from the ribbon icon on the left sidebar

Your API keys are stored locally on your device. They are never sent anywhere except directly to the official OpenAI and Anthropic endpoints.

---

📖 Usage

Starting a chat

Click the AI Vault icon in the left ribbon — the chat panel opens on the right.

Switching models

Click the model name in the chat header to switch between GPT and Claude variants without leaving the conversation.

Using vault context (RAG)

Toggle the vault context button to let the AI search your notes and canvases for relevant information before answering.

Creating a project

Use the Projects menu in the chat header to create a new project, set its system prompt, and assign conversations to it.

Changing mode

Pick Code, Learn, or Thinking mode from the mode selector — the AI will adapt its style accordingly.

---

🔒 Privacy

• API keys are stored locally on your device.
• Conversation history is stored locally, outside your vault.
• RAG index is stored locally, outside your vault.
• Vault content is sent to the model only when you enable vault context, and only the relevant chunks are transmitted.
• No data is sent to any third party other than the official OpenAI / Anthropic APIs.

---

💰 Cost

The plugin itself is free and open source.

You pay only for what you use via your own OpenAI / Anthropic API keys, billed directly by the providers. The built-in cost tracker shows estimated spend per conversation.

---

🛠️ Tech

• Written in TypeScript
• Built with esbuild
• Uses official OpenAI and Anthropic REST APIs with streaming (SSE)
• Local-first architecture — no external servers

---

🐛 Reporting issues

Found a bug or have a feature request? Open an issue on GitHub Issues.

When reporting bugs, please include:

• Obsidian version
• Operating system
• Plugin version
• Steps to reproduce

---

🤝 Contributing

Pull requests are welcome. For larger changes, please open an issue first to discuss what you’d like to change.

---

📜 License

MIT © 2026 JamJan05