Export Explained
Information capsules that bridge the gap between human data and AI agent understanding
The Copy-Paste Problem
❌ Traditional Workflow
1. User opens website/app
2. Selects all content (Ctrl+A)
3. Copies to clipboard
4. Pastes into ChatGPT/Claude
5. Loses structure, context, metadata
6. Agent gets confused, hallucinates
7. User repeats process...
✅ Export Capsule Workflow
1. User clicks Export button
2. System creates structured capsule
3. Agent receives complete context
4. Metadata, trust level included
5. Perfect data preservation
6. Agent processes efficiently
7. ✨ Magic happens
What is an Information Capsule?
An information capsule is a standardized way to package any data with its context, metadata, and trust information so AI agents can understand it perfectly.
Think of it as "copy-paste on steroids" — but instead of losing all structure and context, everything is preserved in a machine-readable format.
Structure
Organized data that machines can parse reliably
Context
Where it came from, why it exists, how to use it
Trust
Cryptographic signatures and verification
Universal Application Sources
Export feeds aren't just for websites. Any application can create information capsules:
Web Apps
Dashboards, docs, user interfaces
Desktop
Documents, databases, project files
Mobile
User data, settings, conversations
CLI Tools
Logs, reports, system info
The Universal Pattern
Any Data Source → Export Feed → LLM understands context
Database Query → Export Feed → Agent processes results
User Conversation → Export Feed → Transfer to another agent
System Logs → Export Feed → AI analyzes patterns
Three Generation Modes
Static
Pre-generated files stored anywhere
Pre-screened content
Works offline
Frozen at build time
Dynamic
Generated on-demand via API
Real-time classification
Personalized exports
Requires backend
Live
Extracted from running application
Interactive consent
Always current
Requires browser
Security & Data Classification
Export feeds automatically scan for sensitive data and apply appropriate security measures.
🔴 CRITICAL
• API keys (sk_, pk_)
• Passwords
• Private keys
• Session tokens
→ Automatically redacted
🟡 SENSITIVE
• Email addresses
• Phone numbers
• Internal URLs
• User IDs
→ User consent required
🟢 PUBLIC
• Documentation
• Marketing content
• Public APIs
• General info
→ Normal export
Security Workflow
1. Scan
Automatic detection
2. Classify
Risk categorization
3. Consent
User chooses
4. Export
Secure capsule
Export Feed Structure
Every export feed follows a standardized structure that agents can reliably parse:
{
"feed_type": "export",
"metadata": {
"title": "Project Documentation",
"origin": "https://myapp.com/docs",
"generated_at": "2025-06-19T10:30:00Z",
"description": "Complete project documentation with examples"
},
"data_classification": {
"security_scan_performed": true,
"sensitive_data_handling": "user_consented",
"classification_level": "public"
},
"content": {
"documentation": "Complete guide to using our API...",
"examples": ["curl -X GET /api/users", "POST /api/login"],
"metadata": {
"word_count": 2847,
"last_updated": "2025-06-15",
"contributors": ["dev-team"]
}
},
"trust": {
"signed_blocks": ["metadata", "content"],
"trust_level": "certified",
"certifier": "https://myapp.com/.well-known/public.pem"
}
}Bundle Archives (.zip)
For complex exports, you can package multiple files into a single archive with the .llmfeed.json acting as the index and descriptor.
{
"feed_type": "export",
"metadata": {
"title": "Complete Project Bundle",
"origin": "https://github.com/user/project"
},
"data": {
"files": [
{
"path": "README.md",
"description": "Project overview",
"security_classification": "public"
},
{
"path": "src/config.js",
"description": "Configuration file",
"security_classification": "sensitive",
"redaction_applied": "credentials_masked"
},
{
"path": "docs/api.pdf",
"description": "API documentation",
"security_classification": "public"
}
],
"security_summary": {
"total_files": 3,
"public_files": 2,
"sensitive_files": 1
}
}
}📦 Use Cases for Bundles
• API provider: credentials.llmfeed.json + usage_guide.md + samples.json
• Researcher: dataset_info.llmfeed.json + data.csv + license.txt + paper.pdf
• Developer: project_context.llmfeed.json + source_code/ + docs/ + assets/
• Enterprise: compliance_report.llmfeed.json + audit_trail.xlsx + certificates/
Why This Matters for Business
Competitive Advantage
• 2010: Websites fought for SEO ranking
• 2025: Websites compete for AIO (Agent Indexing Optimization)
• Early adopters get better agent discovery
• Structured data = more reliable agent interactions
Enterprise Benefits
• Security: Controlled data exposure
• Compliance: Audit trails and consent
• Efficiency: Less copy-paste errors
• Innovation: New agent-driven workflows