Pure Chat LLM

Pure Chat LLM ^_^

Transform your Obsidian notes into powerful AI conversations with built-in agent capabilities.

Installation | Quick Start | Features | Documentation

Overview ?

Pure Chat LLM turns Obsidian notes into interactive AI chat interfaces with unprecedented integration into your knowledge base. Unlike typical AI plugins, Pure Chat LLM provides 13 intelligent tools that allow AI models to autonomously read, write, search, and manage your vault -- with your explicit approval.

What Makes It Different? >>>

• Agentic AI > : AI can autonomously use tools to interact with your vault (with your permission)
• Note-as-Chat = : Transform any note into a conversation -- no separate chat windows
• Deep Integration & : Leverages Obsidian's API for vault operations, workspace management, and more
• Safety First ! : All destructive operations require explicit user approval via visual diff modals
• 42.8% Smaller v : Optimized architecture (120.7KB bundle, down from 211KB in v1.x)
• Voice Calls o : Real-time voice conversations with tool access via OpenAI Realtime API
• Multi-Provider * : OpenAI, Gemini, xAI, Anthropic, Cohere, Mistral AI, DeepSeek

Features +

Intelligent Tool System (Agent Mode)

Pure Chat LLM includes 13 sophisticated tools that enable AI to work with your vault:

Vault Operations =

• Read notes with section/heading/block support ([[Note#Header]], [[Note#^block]])
• Write notes with create/append/prepend modes + user approval modal
• Search content with boolean logic (AND/OR/NOT), regex, and context windows
• Discover files using glob patterns (**/*.md, src/**/*.ts)
• Analyze backlinks to explore your knowledge graph

AI Capabilities *

• Generate images (OpenAI, xAI, Gemini) with aspect ratio control
• Semantic search via Smart Connections plugin integration (RAG)

System Control @

• Manage settings with confirmation modals for safety
• Apply templates programmatically with placeholder replacement

Workspace Management ^

• Get active context: Current file, cursor position, selection
• Manage workspace: Open/close tabs with split pane options
• Show notices: Display toast notifications

Agent Mode must be enabled in settings to allow AI autonomous tool access during conversations.

Chat Features

• Inline conversations: Transform any note into a chat interface
• Multiple AI providers: Switch between OpenAI, Gemini, xAI, Anthropic, and more
• Note linking: Use [[Note name]] to include content from other notes
• Per-note configuration: Customize model, tokens, temperature via JSON block
• System prompts: Modify AI behavior per conversation
• Markdown rendering: Full formatting support (lists, code blocks, tables)
• Streaming responses: Real-time token streaming for supported models

Voice Call Feature

Real-time voice conversations using OpenAI Realtime API with optional tool access:

• WebRTC audio streaming: Direct connection to OpenAI
• Tool integration: When Agent Mode is enabled, AI can use tools during voice calls
• Natural interaction: "Create a note about my meeting", "Search for project deadlines"
• High-quality audio: Echo cancellation, noise suppression, auto-gain control
• Provider extensibility: Designed to support multiple voice providers

Safety & Privacy !

• Explicit approval: All file writes and settings changes require user confirmation
• Visual diff modals: See exactly what will change before accepting
• Local processing: Notes stay in your vault; only API calls go to providers
• Configurable permissions: Control which tools AI can access

Installation v

From Obsidian Community Plugins (Recommended)

1. Open Settings -> Community Plugins
2. Click Browse and search for Pure Chat LLM
3. Click Install, then Enable
4. Configure your API key (see Configuration)

Manual Installation

1. Download the latest release from GitHub Releases
2. Extract main.js, manifest.json, and styles.css to:

   <vault>/.obsidian/plugins/pure-chat-llm/
   

3. Reload Obsidian
4. Enable the plugin in Settings -> Community Plugins

Quick Start >

Basic Chat

1. Create a note in your vault

2. Write your message:

   # role: user
   
   What is the meaning of life?
   

3. Initiate chat with Shift + Enter or command palette -> "Complete chat response"

4. The AI will respond, and you can continue the conversation

Using Note Context

Include content from other notes using WikiLinks:

# role: user

[[Meeting Notes]]

Based on the meeting notes above, create an action items list.

The content of "Meeting Notes" will be included automatically.

With Configuration

Customize per-note with a JSON block at the top:

{
"model": "gpt-4",
"max_completion_tokens": 4096,
"temperature": 0.7
}

# role: system

You are a helpful writing assistant specializing in technical documentation.

# role: user

Help me improve this README section...

Using Tools (Agent Mode)

Enable Agent Mode in settings, then:

# role: user

Search my vault for all notes about "project deadlines" and create a summary note.

The AI will:

1. Use search_vault to find relevant notes
2. Read the content using read_note_section
3. Use write_note_section to create the summary (with your approval)

Configuration @

API Keys

1. Navigate to Settings -> Pure Chat LLM
2. Enter your API key for your chosen provider:
   • OpenAI: Get API Key
   • Gemini: Google AI Studio
   • xAI: xAI Console
   • Anthropic: Anthropic Console
3. Save settings

Endpoints & Models

Configure multiple endpoints for different providers:

• Name: Friendly name (e.g., "OpenAI GPT-4")
• Endpoint: API URL (e.g., https://api.openai.com/v1/chat/completions)
• API Key: Your provider's key
• Models: Available model list for dropdown

Agent Mode

Enable autonomous tool usage:

1. Go to Settings -> Pure Chat LLM
2. Toggle Agent Mode on
3. Select which tools to make available
4. Configure tool-specific settings (e.g., max recursion depth)

Voice Calls

For voice call support:

1. Configure OpenAI API key with Realtime API access
2. Grant microphone permissions
3. Click phone icon in ribbon or use command palette -> "Open voice call panel"

Usage Examples $

Example 1: Research Assistant

{
"model": "gpt-4",
"tools": ["search_vault", "read_note_section", "smart_connections_rag"]
}

# role: system

You are a research assistant. Use the available tools to find relevant information from the user's vault.

# role: user

Find all notes related to "quantum computing" and summarize the key concepts.

Example 2: Note Organizer

{
"model": "gpt-4",
"tools": ["glob_vault_files", "read_note_section", "write_note_section"]
}

# role: user

Review all notes in my "Projects" folder and create an index note with links organized by topic.

Example 3: Writing Coach

{
"model": "claude-3-opus",
"temperature": 0.7
}

# role: system

You are a writing coach. Help improve writing clarity and structure.

# role: user

[[Draft Article]]

Review this draft and suggest improvements for clarity and flow.

Example 4: Voice Meeting Transcription

1. Open voice call panel
2. Start call with Agent Mode enabled
3. Say: "Take notes on our conversation and create a meeting note when we're done"
4. Have your discussion
5. AI creates the note automatically (with your approval)

Available Tools *

See the complete Tool Documentation for detailed information on all 13 tools.

Screenshots o

Chat Interface

Transform any note into an interactive conversation:

# role: user

What are the main themes in my literature review notes?

# role: assistant

I'll search your vault for literature review notes and analyze the themes...

Tool Approval Modal

All destructive operations require explicit approval with diff visualization:

+-------------------------------------+
|  Review Changes                     |
+-------------------------------------+
|  - Old content                      |
|  + New content                      |
|                                     |
|  (Accept)  (Reject)                 |
+-------------------------------------+

Voice Call Panel

Real-time voice conversations with optional tool access:

+-------------------------------------+
|   @  Voice Call                     |
+-------------------------------------+
|  Status: Connected                  |
|  Agent Mode: Enabled                |
|                                     |
|  (~ Mute)  (@ End Call)             |
+-------------------------------------+

Troubleshooting !

Chat Issues

Q: "Complete chat response" does nothing

• Verify API key is configured in settings
• Check your model is available at the endpoint
• Ensure note has proper role formatting (# role: user)

Q: Note links [[Note]] aren't working

• Links must be on their own line
• Verify the linked note exists
• Check recursion depth setting if nested links

Tool Issues

Q: AI says tools aren't available

• Enable Agent Mode in settings
• Verify tools are selected in tool configuration
• Check tool-specific requirements (e.g., Smart Connections plugin)

Q: File write failed

• Ensure you clicked "Accept" in the approval modal
• Check file/folder permissions
• Verify vault path is accessible

Voice Call Issues

Q: Connection fails

• Verify OpenAI API key has Realtime API access
• Check microphone permissions
• Ensure WebRTC is supported (desktop app recommended)

Q: Tools don't work during voice calls

• Enable Agent Mode in settings
• Grant tool permissions
• Check browser console for errors

Architecture ^

v2.0.0 Improvements

Pure Chat LLM v2.0.0 represents a major architectural refactoring:

• 38% bundle size reduction (211KB -> 131KB main.js, plus 84% styles.css reduction)
• Simplified file resolution: Removed complex BlueFileResolver system (~70KB) in favor of lightweight recursive resolution
• 100% JSDoc coverage: All 26 TypeScript files fully documented
• Enhanced type safety: Improved error handling and type definitions
• Code quality tools: Integrated Knip for detecting unused code

Technical Highlights

• Recursive resolution with cycle detection and depth limiting
• Chunked binary encoding for large media files (prevents stack overflow)
• Tool registry pattern for extensible agent capabilities
• Provider abstraction for multi-LLM support
• WebRTC integration for real-time voice (OpenAI Realtime API)

Contributing &

Contributions are welcome! Please:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

Development Setup

# Clone the repository
git clone https://github.com/TheJusticeMan/pure-chat-llm.git
cd pure-chat-llm

# Install dependencies
npm install

# Build for development (watch mode)
npm run dev

# Build for production
npm run build

# Run linting
npm run lint

# Run code quality checks
npm run knip

Documentation i

• Tool Guide: Comprehensive tool documentation (1,662 lines)
• Changelog: Version history and breaking changes
• OpenAI API Docs: API reference
• Obsidian Plugin Dev: Plugin development guide

License ~

MIT License - see LICENSE file for details.

Credits <3

Author: Justice Vellacott Website: thejusticeman.github.io

Acknowledgments

• Obsidian team for the amazing platform
• OpenAI for GPT models and Realtime API
• All contributors and users providing feedback

Links

• GitHub Repository
• Obsidian Community Plugins
• Report Issues
• Feature Requests

Star (*) this repository if you find it helpful!

Made with <3 for the Obsidian community