# Lore DB — Complete Reference

> Supercharge your AI agent with knowledge.

Lore DB is a shared knowledge management system with AI-powered search (RAG), built for teams that use AI coding assistants and want to give them access to internal documentation. Enterprise-grade security, GDPR compliant, zero vendor lock-in.

- Website: https://loredb.dev
- Documentation: https://loredb.dev/docs
- Security & Trust: https://loredb.dev/docs/reference/trust
- Privacy & Data: https://loredb.dev/docs/reference/privacy
- Terms of Service: https://loredb.dev/docs/reference/terms
- Pricing: https://loredb.dev/pricing
- Changelog: https://loredb.dev/changelog
- MCP endpoint: https://loredb.dev/mcp

---

## Security & Trust

Lore DB is built with enterprise-grade security by default. Full details: https://loredb.dev/docs/reference/trust

### Encryption
- AES-256 encryption at rest (Supabase PostgreSQL)
- HTTPS/TLS encryption for all traffic in transit
- API keys encrypted with Fernet symmetric encryption before storage
- Embeddings stored as pgvector arrays in the encrypted database

### OpenAI data policy
- We use the **OpenAI API** (not ChatGPT)
- Per OpenAI's API data usage policy: **your data is never used for model training**
- API data retained for **30 days** for abuse monitoring only, then deleted
- Only document content and queries are sent to OpenAI — never email, name, profile, library names, or account metadata
- Reference: https://openai.com/policies/api-data-usage-policies

### Bring Your Own Key (BYOK)
- Use your own OpenAI API key for a direct contractual relationship with OpenAI
- All AI requests go through your key with your OpenAI data policies
- Configure in Settings within the app

### Authentication
- Google OAuth only — no passwords are stored
- Two-factor authentication via your Google account
- Secure HTTP-only session cookies
- MCP connections use the same authentication system

### Access control
- Role-based: owner, editor, viewer per library
- Systems group libraries for organization-wide access
- Every request is authenticated and authorized — no permission caching
- Community libraries: admin-published only, read-only subscriptions

### GDPR compliance
- Right of access — view all your data in the app
- Right to portability — export any library as JSON
- Right to rectification — edit any document you own
- Right to erasure — delete documents immediately, full account deletion on request
- Right to restriction — toggle libraries inactive to exclude from AI processing
- Data Processing Agreement (DPA) available on request

### Data residency
| Component | Location | Provider |
|-----------|----------|----------|
| Database | EU or US (configurable) | Supabase (PostgreSQL + pgvector) |
| Application | Global edge network | Vercel (serverless, stateless) |
| AI processing | US | OpenAI API |
| Payments | Stripe infrastructure | Stripe |
| Authentication | Google infrastructure | Google |

### What we do NOT do
- We do NOT store passwords (Google OAuth only)
- We do NOT use your data to train AI models
- We do NOT share your documents with other users (role-based access only)
- We do NOT log document content (only metadata: operation type, token counts, timestamps)
- We do NOT sell your data to third parties
- We do NOT allow unauthorized access — every request is authenticated

### Data retention
| Data | Retained until |
|------|---------------|
| Documents | You delete them |
| Embeddings | Regenerated on edit, deleted with document |
| Usage logs | Indefinitely (metadata only, no content) |
| User accounts | You request deletion |
| OpenAI API data | 30 days (per OpenAI policy) |

### Zero vendor lock-in
- Export any library as complete JSON (titles, content, metadata)
- Documents stored as standard Markdown
- Embeddings use standard OpenAI text-embedding-3-small format
- No proprietary data formats

---

## Product Overview

Lore DB lets teams store documents in Markdown, organize them into shared libraries, and connect them to AI tools via the Model Context Protocol (MCP). When connected, AI assistants like Claude, Cursor, VS Code, ChatGPT, Claude.ai, and Gemini CLI can search, read, and reason over your team's real documentation — giving grounded answers instead of hallucinations.

### Core capabilities

1. **Document storage** — Write in Markdown with full rendering. Documents are automatically chunked and embedded for semantic search.
2. **Document linking** — Use `[[Document Title]]` wiki syntax to create links between documents. Linked documents and backlinks are shown on each document view.
3. **Knowledge graph** — Interactive force-directed visualization of all documents and their connections. Click nodes to navigate.
4. **AI-powered Q&A** — Ask questions in natural language, get answers with source citations from your actual documents. Linked documents are automatically pulled in as extra context.
5. **Hybrid search** — Combines semantic similarity, keyword matching, and freshness ranking for the best results.
6. **Shared libraries** — Organize documents into collections with owner/editor/viewer roles. Share via invite links.
7. **Community libraries** — Browse and subscribe to publicly shared knowledge bases. Admin-curated, read-only access.
8. **MCP server** — First-class Model Context Protocol support. Any MCP-compatible AI tool can access your knowledge base.
9. **Import/export** — Libraries can be exported as portable JSON and imported elsewhere. Zero lock-in.
10. **Search controls** — Toggle which libraries feed into AI answers per user.

---

## MCP Integration

### Connecting Claude Desktop

Add to your Claude Desktop MCP configuration:

```json
{
  "mcpServers": {
    "lore-db": {
      "url": "https://loredb.dev/mcp"
    }
  }
}
```

### Connecting Claude Code (CLI)

```bash
claude mcp add lore-db --transport http https://loredb.dev/mcp
```

### Connecting Cursor / VS Code

Add to `.cursor/mcp.json` or VS Code MCP settings:

```json
{
  "mcpServers": {
    "lore-db": {
      "url": "https://loredb.dev/mcp"
    }
  }
}
```

No plugins, packages, or local dependencies required. Lore DB serves MCP over HTTP directly.

### Available MCP tools

When connected, the following tools are available to the AI agent:

- `search_docs` — Search documents by query (hybrid semantic + keyword search)
- `ask` — Ask a question and get an AI-generated answer with source citations
- `read_doc` — Read a specific document by ID
- `create_doc` — Create a new document with title and Markdown content
- `update_doc` — Update an existing document
- `delete_doc` — Delete a document
- `list_docs` — List all accessible documents
- `list_libraries` — List all libraries the user has access to

---

## Use Cases

### Engineering teams
Give AI coding assistants (Cursor, Claude Code, VS Code) access to your internal API docs, architecture decisions, deployment runbooks, and coding standards. Your AI gives answers grounded in your real documentation. Guide: https://loredb.dev/docs/use-cases/developers

### Support teams
Let AI agents search your knowledge base to draft accurate customer responses. Reduce time spent looking up internal documentation. Guide: https://loredb.dev/docs/use-cases/support-teams

### Team knowledge sharing
Build a shared source of truth. New team members ask their AI assistant questions and get answers from real company documentation instead of outdated wikis. Guide: https://loredb.dev/docs/use-cases/team-knowledge

---

## Technical Architecture

- **Backend**: Python with FastMCP framework, deployed as a Vercel serverless function (stateless)
- **Database**: Supabase PostgreSQL with pgvector extension for vector similarity search, AES-256 encrypted
- **Embeddings**: OpenAI text-embedding-3-small (1536 dimensions)
- **Chat models**: Configurable (default: GPT-4o), with per-model input/output token pricing
- **Frontend**: Astro with React islands, Tailwind CSS v4, TanStack Router
- **Authentication**: Google OAuth 2.0 with secure HTTP-only session cookies
- **Search**: Hybrid approach — cosine similarity on embeddings + trigram keyword matching + freshness boost
- **Infrastructure**: Vercel serverless (stateless, no persistent server), Supabase (managed PostgreSQL)

---

## Library Roles & Access Control

| Role | Permissions |
|------|------------|
| Owner | Full control: edit library, manage members, add/remove documents, delete library, generate invite links |
| Editor | Add and remove documents from the library |
| Viewer | View and search documents in the library (read-only) |

Systems group multiple libraries. A system membership grants access to all libraries within that system.

Full access control documentation: https://loredb.dev/docs/reference/access-control

---

## Pricing

Flat-rate plans with weekly usage limits on Ask AI. Search and document processing use local embeddings and are free on all plans.

| Plan | Price | Documents | Libraries | Ask AI |
|------|-------|-----------|-----------|--------|
| Free | SEK 0/month | 25 | 1 | Weekly limit |
| Pro | SEK 79/month | Unlimited | Unlimited | Unlimited |

- BYOK: Use your own OpenAI API key to bypass usage limits entirely
- Bonus credits can be granted by admins for additional Ask AI overflow
- Free to try, no credit card required
- Cancel anytime, 30 days notice for pricing changes

---

## Community Libraries

Community libraries are published by Lore DB administrators and contain freely available public content (government regulations, public guidelines, open-source documentation). Any user can subscribe for read-only access.

- Only administrators can publish to the community
- No user-uploaded copyrighted content is distributed
- Users can create private libraries with their own content
- Browse at: https://loredb.dev/community

---

## Comparisons

- Notion vs Lore DB: https://loredb.dev/docs/compare/notion
- Confluence vs Lore DB: https://loredb.dev/docs/compare/confluence
- Glean vs Lore DB: https://loredb.dev/docs/compare/glean
- Build Your Own vs Lore DB: https://loredb.dev/docs/compare/build-your-own

---

## Documentation Index

Complete documentation: https://loredb.dev/docs

### Getting started
- Getting started guide: https://loredb.dev/docs/getting-started

### Integration guides
- Connect Claude Desktop: https://loredb.dev/docs/guides/connect-claude-desktop
- Connect Claude Code: https://loredb.dev/docs/guides/connect-claude-code
- Connect Cursor: https://loredb.dev/docs/guides/connect-cursor
- Connect ChatGPT: https://loredb.dev/docs/guides/connect-chatgpt
- Connect other tools: https://loredb.dev/docs/guides/connect-other

### Usage guides
- Share with your team: https://loredb.dev/docs/guides/share-with-team
- Organize libraries: https://loredb.dev/docs/guides/organize-libraries
- Setup systems: https://loredb.dev/docs/guides/setup-systems
- Content tips: https://loredb.dev/docs/guides/content-tips
- Team rollout: https://loredb.dev/docs/guides/team-rollout
- Import & export: https://loredb.dev/docs/guides/import-export
- Data portability: https://loredb.dev/docs/guides/data-portability

### Reference
- Security & Trust: https://loredb.dev/docs/reference/trust
- Security details: https://loredb.dev/docs/reference/security
- Privacy & Data: https://loredb.dev/docs/reference/privacy
- Terms of Service: https://loredb.dev/docs/reference/terms
- MCP Tools: https://loredb.dev/docs/reference/mcp-tools
- Access Control: https://loredb.dev/docs/reference/access-control
- Search: https://loredb.dev/docs/reference/search
- Ask AI: https://loredb.dev/docs/reference/ask-ai
- Libraries: https://loredb.dev/docs/reference/libraries
- Systems: https://loredb.dev/docs/reference/systems
- Documents: https://loredb.dev/docs/reference/documents
- Billing: https://loredb.dev/docs/reference/billing
- REST API: https://loredb.dev/docs/reference/api

### Prompts & Skills
- Copy-paste prompts: https://loredb.dev/docs/prompts
- Workflow templates: https://loredb.dev/docs/skills