Feature Request: Integrate AI conversation capability with context awareness

[kookyleo]

Feature Description

Add AI conversation capability to Markon with context awareness, allowing users to interact with AI assistants while viewing/annotating markdown documents.

Use Cases

1. Document Q&A: Ask questions about the current document content
2. Annotation Assistance: Get AI help for summarizing, explaining, or expanding selected text
3. Writing Support: Request AI assistance for note-taking, rephrasing, or translation
4. Context-Aware Chat: Maintain conversation history with document context

Proposed Integration Approaches

Option 1: Sidebar Chat Panel

• Add a collapsible chat panel (similar to annotation sidebar)
• Toggle with keyboard shortcut (e.g., `Ctrl+Shift+A`)
• Include current document context in AI requests
• Support markdown rendering in chat responses

Option 2: Selection-Based Interaction

• Right-click or keyboard shortcut on selected text
• Quick AI actions: "Summarize", "Explain", "Translate", "Expand"
• Results displayed as annotations or in a modal

Option 3: Command Palette

• Press `/` or `Ctrl+K` to open command palette
• Type AI commands like `/ask`, `/summarize`, `/translate`
• Send selected text or full document as context

Technical Considerations

Backend Options

1. OpenAI API Integration

   • Support for GPT-4, GPT-3.5-turbo
   • Configurable API key (environment variable or settings)
   • Streaming responses for better UX

2. Anthropic Claude API

   • Claude 3 Opus/Sonnet/Haiku support
   • Better for long documents (200K context window)

3. Local LLM Support

   • Ollama integration
   • LM Studio compatibility
   • Privacy-friendly for sensitive documents

4. Pluggable Architecture

   • Provider abstraction layer
   • Support multiple AI backends
   • Allow custom endpoints

Context Management

```rust
struct AIContext {
document_path: String,
document_content: String, // Full or relevant sections
selected_text: Option,
conversation_history: Vec,
annotations: Vec, // Existing annotations as context
}
```

Configuration

```toml

~/.config/markon/config.toml or .markon.toml

[ai]
enabled = true
provider = "openai" # openai, anthropic, ollama, custom
api_key = "sk-..." # Or use MARKON_AI_API_KEY env var
model = "gpt-4"
max_context_tokens = 8000
temperature = 0.7

[ai.shortcuts]
toggle_chat = "Ctrl+Shift+A"
quick_ask = "Ctrl+Shift+Q"
summarize_selection = "Ctrl+Shift+S"
```

Privacy & Security

• API keys stored securely (system keychain or env vars)
• Option to disable AI for sensitive documents
• Local processing option (Ollama/LM Studio)
• Explicit user consent before sending data
• Configurable data retention policy

UI/UX Features

Chat Interface

• Markdown rendering for responses
• Code syntax highlighting in AI responses
• Copy response to clipboard
• Insert response as annotation
• Export conversation history

Visual Indicators

• Show when AI is processing (loading spinner)
• Token usage display (for API-based providers)
• Context length indicator
• Error handling with user-friendly messages

Implementation Phases

Phase 1: Basic Integration (MVP)

• OpenAI API integration
• Simple chat sidebar
• Send selected text or full document as context
• Basic configuration (API key)

Phase 2: Enhanced Features

• Multiple AI provider support
• Context-aware suggestions
• Annotation integration (AI-generated notes)
• Conversation history persistence

Phase 3: Advanced Features

• Local LLM support (Ollama)
• Custom prompts/templates
• Multi-document context
• Collaborative AI sessions (shared mode)

Example User Workflows

Workflow 1: Document Understanding

1. Open a complex markdown file
2. Press `Ctrl+Shift+A` to open AI chat
3. Ask: "What is the main topic of this document?"
4. AI provides summary with document context

Workflow 2: Annotation Enhancement

1. Select technical jargon in document
2. Press `Ctrl+Shift+S` for quick summarize
3. AI explains the term in simple language
4. Save explanation as annotation

Workflow 3: Writing Assistant

1. Writing notes in annotation
2. Ask AI: "Rephrase this more professionally"
3. AI provides alternatives
4. Choose and apply to annotation

Related Tools for Inspiration

• Obsidian AI plugins
• Notion AI
• GitHub Copilot Chat
• Continue.dev - AI assistant for IDEs

Open Questions

1. Default provider: Should we recommend one provider, or leave it completely open?
2. Pricing concerns: Should we include usage cost estimates?
3. Offline mode: Priority for local LLM support?
4. Multi-language: Should AI respect document language for responses?
5. Streaming UI: Real-time streaming vs. complete response?

Community Input Welcome

This is a significant feature addition. Community feedback on:

• Preferred AI providers
• Most valuable use cases
• Privacy concerns
• UI/UX preferences
• Configuration complexity tolerance

---

Priority: Medium (nice-to-have, not critical)
Complexity: High (requires significant backend and frontend work)