> ## Documentation Index
> Fetch the complete documentation index at: https://docs.plugged.in/llms.txt
> Use this file to discover all available pages before exploring further.

# Version 2.8.0 - AI Document Exchange

> Revolutionary document management with AI model attribution

# Version 2.8.0 Release Notes

<Badge variant="info">Feature Release</Badge>
<Badge variant="success">RAG v2</Badge>

Released: January 27, 2025

## Overview

Version 2.8.0 introduces the **AI Document Exchange** system (RAG v2), a revolutionary approach to document management that enables AI models to create, manage, and collaborate on documents with full attribution tracking.

## Key Features

<CardGroup cols={2}>
  <Card title="AI Document Creation" icon="robot">
    Direct document generation by AI models with attribution
  </Card>

  <Card title="Version Control" icon="code-branch">
    Complete version history with change tracking
  </Card>

  <Card title="Semantic Search" icon="magnifying-glass">
    Advanced search with relevance scoring
  </Card>

  <Card title="Multi-Format Support" icon="file-code">
    Support for Markdown, JSON, HTML, and more
  </Card>
</CardGroup>

## AI Document Exchange System

### Document Creation API

AI models can now create documents directly through the API:

```bash theme={null}
POST /api/documents/ai
Authorization: Bearer YOUR_API_KEY

{
  "title": "MCP Integration Guide",
  "content": "# Complete guide to MCP integration...",
  "format": "md",
  "metadata": {
    "model": {
      "name": "Claude 3",
      "provider": "Anthropic",
      "version": "3.0"
    },
    "context": "User requested comprehensive MCP guide",
    "visibility": "workspace"
  },
  "tags": ["mcp", "integration", "guide"]
}
```

### Model Attribution

Every AI-generated document tracks complete attribution:

```json theme={null}
{
  "document_id": "doc_abc123",
  "attributions": [
    {
      "model_name": "Claude 3",
      "model_provider": "Anthropic",
      "model_version": "3.0",
      "contribution_type": "created",
      "timestamp": "2025-01-27T10:00:00Z",
      "metadata": {
        "tokens_used": 1500,
        "confidence_score": 0.95
      }
    },
    {
      "model_name": "GPT-4",
      "model_provider": "OpenAI",
      "contribution_type": "reviewed",
      "timestamp": "2025-01-27T11:00:00Z"
    }
  ]
}
```

## Document Management Features

### Version Control

<Info>
  All documents now support complete version history with rollback capabilities.
</Info>

#### Version Tracking

* Automatic version creation on updates
* Parent-child relationships between versions
* Change diff generation
* Rollback to any previous version

#### Version API

```typescript theme={null}
// Get version history
GET /api/documents/{id}/versions

// Create new version
POST /api/documents/{id}/versions
{
  "content": "Updated content",
  "changeSummary": "Fixed typos and added examples"
}

// Rollback to version
POST /api/documents/{id}/rollback/{versionId}
```

### Advanced Search

PostgreSQL full-text search with relevance scoring:

```typescript theme={null}
POST /api/documents/search
{
  "query": "MCP server configuration",
  "filters": {
    "modelName": "Claude 3",
    "modelProvider": "Anthropic",
    "dateFrom": "2025-01-01",
    "dateTo": "2025-01-31",
    "source": "ai_generated",
    "tags": ["mcp", "configuration"]
  },
  "limit": 10,
  "offset": 0
}
```

**Search Features:**

* Semantic relevance scoring
* Automatic snippet generation
* Keyword highlighting
* Filter by AI model, date, tags, source
* Pagination with total count

## Document Preview System

### Multi-Format Viewer

The new document preview modal supports:

<Tabs>
  <Tab title="PDF">
    * Page navigation
    * Zoom controls (10%-500%)
    * Text selection
    * Search within document
  </Tab>

  <Tab title="Images">
    * Pan and zoom
    * Rotation controls
    * Fullscreen mode
    * Format support: PNG, JPG, GIF, WebP, SVG
  </Tab>

  <Tab title="Text/Code">
    * Syntax highlighting for 20+ languages
    * Line numbers
    * Code folding
    * Copy to clipboard
  </Tab>

  <Tab title="Markdown">
    * Rendered preview
    * Table of contents
    * Link navigation
    * Embedded media
  </Tab>
</Tabs>

### Preview Controls

```typescript theme={null}
// Preview component usage
<DocumentPreview
  document={document}
  onClose={() => setPreviewOpen(false)}
  onNavigate={(direction) => navigateDocument(direction)}
  fullscreen={true}
/>
```

## MCP Tools Integration

New MCP tools for document operations:

### pluggedin\_create\_document

Create documents from MCP servers:

```json theme={null}
{
  "name": "pluggedin_create_document",
  "parameters": {
    "title": "API Documentation",
    "content": "# API Reference...",
    "format": "md",
    "category": "documentation",
    "tags": ["api", "reference"]
  }
}
```

### pluggedin\_search\_documents

Search documents semantically:

```json theme={null}
{
  "name": "pluggedin_search_documents",
  "parameters": {
    "query": "authentication flow",
    "limit": 5
  }
}
```

### pluggedin\_get\_document

Retrieve specific documents:

```json theme={null}
{
  "name": "pluggedin_get_document",
  "parameters": {
    "documentId": "doc_abc123",
    "includeContent": true,
    "includeVersions": false
  }
}
```

### pluggedin\_update\_document

Update existing documents:

```json theme={null}
{
  "name": "pluggedin_update_document",
  "parameters": {
    "documentId": "doc_abc123",
    "operation": "append",
    "content": "\n## New Section\nAdditional content..."
  }
}
```

## Security Enhancements

### Rate Limiting

* AI document creation: 10 requests per hour
* Document search: 60 requests per minute
* Document updates: 30 requests per minute

### Content Security

<Warning>
  All AI-generated content is sanitized before storage.
</Warning>

* HTML/Markdown sanitization
* Path traversal protection
* Content size limit: 10MB
* File type validation
* XSS prevention

### Access Control

```typescript theme={null}
// Document visibility levels
enum Visibility {
  PRIVATE = 'private',      // Only owner
  WORKSPACE = 'workspace',   // Project members
  PUBLIC = 'public'         // Everyone
}
```

## Database Schema Updates

### New Tables

#### document\_versions

```sql theme={null}
CREATE TABLE document_versions (
  id UUID PRIMARY KEY,
  document_id UUID REFERENCES docs(id),
  version_number INTEGER,
  content TEXT,
  content_hash VARCHAR(64),
  change_summary TEXT,
  created_at TIMESTAMP,
  created_by UUID REFERENCES users(id)
);
```

#### document\_model\_attributions

```sql theme={null}
CREATE TABLE document_model_attributions (
  id UUID PRIMARY KEY,
  document_id UUID REFERENCES docs(id),
  model_name VARCHAR(255),
  model_provider VARCHAR(255),
  model_version VARCHAR(50),
  contribution_type VARCHAR(50),
  metadata JSONB,
  created_at TIMESTAMP
);
```

### Updated docs Table

New fields added:

* `source`: 'upload' | 'ai\_generated' | 'api'
* `ai_metadata`: JSONB for AI-specific data
* `content_hash`: SHA-256 hash for deduplication
* `visibility`: Document visibility level
* `version`: Current version number
* `parent_document_id`: For version relationships

## UI Enhancements

### Library Interface

<Steps>
  <Step title="Source Filter">
    Filter documents by source (Uploaded, AI Generated, API)
  </Step>

  <Step title="Model Attribution Badges">
    Visual indicators for AI-generated content
  </Step>

  <Step title="Progress Tracking">
    Real-time upload and processing progress
  </Step>

  <Step title="Preview on Click">
    Instant document preview in modal
  </Step>
</Steps>

### Upload Progress

Enhanced progress tracking shows:

* Text extraction progress
* Chunking progress
* Embedding generation
* Database insertion
* Estimated time remaining

## Migration Guide

### Database Migration

Run the migration script:

```bash theme={null}
# Run new migrations
pnpm db:migrate

# The migration adds:
# - document_versions table
# - document_model_attributions table
# - New fields to docs table
```

### API Updates

Update your API calls to include model metadata:

```javascript theme={null}
// Old API call
await createDocument({
  title: "My Document",
  content: "Content here"
});

// New API call with attribution
await createDocument({
  title: "My Document",
  content: "Content here",
  metadata: {
    model: {
      name: "Claude 3",
      provider: "Anthropic",
      version: "3.0"
    }
  }
});
```

## Performance Improvements

* Optimized search queries with new indexes
* Lazy loading for document previews
* Content deduplication reduces storage by 30%
* Streaming support for large documents

## Known Issues

* PDF preview may be slow for documents >100 pages
* Search relevance scoring needs tuning for short queries
* Version diff generation slow for very large documents

## Future Enhancements

Planned for future releases:

* Real-time collaborative editing
* Document templates
* Advanced version merging
* AI-powered document summarization
* Export to multiple formats

## Support

For help with RAG v2 features:

* **Documentation**: [docs.plugged.in](https://docs.plugged.in)
* **API Reference**: [API Documentation](/api/reference)
* **GitHub Issues**: [Report issues](https://github.com/VeriTeknik/pluggedin-app/issues)