Connectors

​

Supported platforms

​

How connectors work

1. OAuth authentication - You authorize Scira to access your content
2. Document indexing - Your documents are synced to Supermemory
3. Semantic processing - Content is chunked and embedded for AI search
4. Isolated storage - Your data is tagged with your unique user ID

​

Search process

// Scira's connector search tool
const result = await client.search.documents({
  q: query,
  containerTags: [userId, config.syncTag],
  limit: 15,
  rerank: true,
  includeSummary: true,
});

• Uses semantic similarity to find relevant documents
• Re-ranks results for accuracy
• Returns document chunks with relevance scores
• Generates direct links to source documents

​

Connecting a service

​

Google Drive

1. Navigate to your Scira Pro settings
2. Click Connect Google Drive
3. Authorize Scira to access your Drive files
4. Wait for initial sync (may take a few minutes)

// Connection configuration
const config = {
  name: 'Google Drive',
  documentLimit: 3000,
  syncTag: 'gdrive-sync',
};

​

Notion

1. Navigate to your Scira Pro settings
2. Click Connect Notion
3. Select the pages/databases to share with Scira
4. Authorize the integration
5. Wait for initial sync

// Connection configuration
const config = {
  name: 'Notion',
  documentLimit: 2000,
  syncTag: 'notion-workspace',
};

​

OneDrive (coming soon)

​

Using connectors

"Find my meeting notes from last week"
"What did I write about project planning?"
"Show me documents related to marketing strategy"

​

Implementation details

​

Document metadata

interface EnhancedDocument {
  documentId: string;
  title: string | null;
  content: string | null;
  summary: string | null;
  chunks: Array<{
    content: string;
    score: number;
    isRelevant: boolean;
  }>;
  score: number;
  metadata: Record<string, unknown> | null;
  provider: 'google-drive' | 'notion' | 'onedrive' | null;
  url: string; // Direct link to source
  type: string | null; // Document type/mime type
  createdAt: string;
  updatedAt: string;
}

​

URL generation

• Google Drive: https://drive.google.com/file/d/{documentId}/view
• Notion: https://notion.so/{title-slug}-{documentId}
• OneDrive: https://1drv.ms/b/s!{documentId}

​

Content filtering

// Documents must have valid content
const hasValidContent = (doc) => {
  // Must have non-empty chunks, summary, or content
  // Excludes "Empty Chunk" placeholders
  return doc.chunks?.some(chunk => 
    chunk.content?.trim() && 
    chunk.content !== 'Empty Chunk'
  );
};

​

Privacy and security

​

Data isolation

const containerTags = [
  userId,              // Your unique user ID
  config.syncTag,      // Service identifier
];

​

What gets synced

• Text content - Extracted and chunked for search
• Metadata - Title, type, creation/update dates
• Source links - References back to original documents

​

What doesn’t get synced

• File attachments - Only text content is indexed
• Images - Visual content is not processed
• Private/restricted - Only documents you explicitly authorize

​

Managing connections

​

View connection status

const status = await getSyncStatus(provider, userId);
// Returns:
// {
//   isConnected: true,
//   documentCount: 1247,
//   lastSync: "2024-01-15T10:30:00Z",
//   email: "[email protected]",
//   status: "active"
// }

​

Manual sync

await manualSync(provider, userId);

​

Disconnect a service

await deleteConnection(connectionId);

​

Supermemory integration

• Semantic search - Understanding meaning, not just keywords
• Automatic chunking - Intelligent document segmentation
• Re-ranking - Results sorted by relevance
• Incremental sync - Only new/changed documents are processed

​

API configuration

import Supermemory from 'supermemory';

const client = new Supermemory({
  apiKey: process.env.SUPERMEMORY_API_KEY,
});

​

Troubleshooting

​

No results found

• Check sync status - Ensure documents have been indexed
• Verify permissions - Confirm Scira has access to the documents
• Try broader queries - Use semantic queries rather than exact phrases

​

Sync issues

• Document limits - You may have exceeded the provider’s limit
• API errors - Check your Supermemory API key is valid
• OAuth expiry - Reconnect if authorization has expired

​

Performance

• Initial sync - First sync may take several minutes for large accounts
• Search speed - Typically returns results in 1-3 seconds
• Re-ranking - Improves accuracy but adds slight latency

​

Best practices

1. Organize your content - Well-structured documents search better
2. Use descriptive titles - Helps with semantic matching
3. Regular syncs - Keep your index up to date
4. Specific queries - More context leads to better results
5. Monitor limits - Stay within document count restrictions

​

API reference

• createConnection(provider, userId) - Initialize OAuth flow
• getConnection(provider, userId) - Check connection status
• listUserConnections(userId) - Get all user connections
• deleteConnection(connectionId) - Remove connection
• manualSync(provider, userId) - Trigger document sync
• getSyncStatus(provider, userId) - Get sync statistics

• createConnectorsSearchTool(userId, selectedConnectors) - Create search tool instance