MinKy - Team Knowledge Intelligence Platform

Turn tacit knowledge into searchable, connected team intelligence

Features | Quick Start | Architecture | API Reference | Documentation | Contributing

---

Overview

MinKy is a knowledge intelligence platform designed for small teams (3-9 members) that transforms tacit knowledge into explicit, searchable assets. Unlike traditional document management systems that rely on manual tagging and categorization, MinKy uses AI to understand, connect, and surface relevant knowledge through natural language conversations.

The Problem

• Team knowledge is scattered across Slack, emails, documents, and individual minds
• Manual tagging is inconsistent and rarely maintained
• Finding relevant past decisions or solutions requires knowing where to look
• When team members leave, their knowledge leaves with them

The Solution

MinKy captures knowledge from various sources, uses AI to understand context and relationships, stores it with vector embeddings, and enables natural language search through RAG (Retrieval-Augmented Generation).

---

Features

Phase 1: Knowledge Understanding (Current)

• AI Document Analysis: Claude analyzes uploaded documents to extract topics, summaries, and insights
• Vector Embeddings: Documents stored with pgvector for semantic similarity search
• Automatic Connections: Related documents linked based on content similarity, not manual tags

Phase 2: Conversational Search (In Progress)

• Natural Language Q&A: Ask questions in plain language, get answers with source citations
• RAG-Powered Search: Combines vector search with AI generation for accurate responses
• Context-Aware Responses: Understands your team's terminology and context

Phase 3: Knowledge Connections (Planned)

• Knowledge Graph: Visualize relationships between documents and concepts
• Smart Recommendations: "People who found this useful also looked at..."
• Gap Detection: Identify undocumented areas in your knowledge base

Additional Features

• Multi-Source Ingestion: Markdown, Safari Clipper, Slack messages
• Korean Language Support: Full Korean NLP with MeCab integration
• OCR Processing: Extract text from images and scanned documents
• Real-Time Collaboration: WebSocket-based collaborative editing
• Analytics Dashboard: Track knowledge usage and team engagement
• Document Workflows: Approval processes with customizable templates
• Version Control: Full document history with diff and restore

---

Quick Start

Prerequisites

• Rust 1.75+ (for backend)
• Node.js 18+ (for frontend)
• PostgreSQL 15+ with pgvector extension
• OpenSearch 2.x (optional, for full-text search)

1. Clone the Repository

git clone https://github.com/hephaex/minky.git
cd minky

2. Set Up the Database

# Create PostgreSQL database
createdb minky
# Enable pgvector extension
psql minky -c "CREATE EXTENSION vector;"

3. Configure Environment

# Copy example configuration
cp minky-rust/.env.example minky-rust/.env
# Edit with your settings
nano minky-rust/.env

Key configuration options:

DATABASE_URL=postgres://user:password@localhost:5432/minkyJWT_SECRET=your-secure-secret-keyANTHROPIC_API_KEY=sk-ant-...# For AI featuresOPENAI_API_KEY=sk-...# For embeddings

4. Run Database Migrations

cd minky-rust
cargo install sqlx-cli
sqlx migrate run

5. Start the Backend

cd minky-rust
cargo run
# Server starts at http://localhost:8000

6. Start the Frontend

cd frontend
npm install
npm start
# App opens at http://localhost:3000

7. Verify Installation

# Check backend health
curl http://localhost:8000/health
# Expected response:# {"status":"healthy","version":"0.1.0"}

For detailed setup instructions, see Docs/GETTING_STARTED.md

---

Architecture

+-------------------+ +-------------------+ +-------------------+
| Frontend | | Rust Backend | | PostgreSQL |
| (React) |<--->| (Axum) |<--->| + pgvector |
| | | | | |
+-------------------+ +--------+----------+ +-------------------+
|
+--------------+--------------+
| | |
+-----v----+ +------v-----+ +-----v-----+
| Claude | | OpenSearch | | OpenAI |
| (AI) | | (Search) | | Embedding |
+----------+ +------------+ +-----------+

Technology Stack

Layer	Technology	Purpose
Frontend	React 18	Single-page application
Backend	Rust + Axum	High-performance API server
Database	PostgreSQL	Primary data storage
Vector DB	pgvector	Semantic embeddings
Search	OpenSearch	Full-text search
AI	Claude API	Document understanding, Q&A
Embedding	OpenAI API	Vector generation

Key Components

• Document Pipeline: Ingest -> AI Analysis -> Embed -> Store -> Index
• RAG Search: Query -> Vector Search -> Context Assembly -> AI Generation
• Real-time: WebSocket for collaborative editing

For detailed architecture, see Docs/ARCHITECTURE.md

---

Project Structure

minky/
├── minky-rust/ # Rust backend (Active)
│ ├── src/
│ │ ├── models/ # Data models (40+ types)
│ │ ├── routes/ # API endpoints (25+ routes)
│ │ ├── services/ # Business logic (20+ services)
│ │ └── middleware/ # Auth, rate limiting
│ └── migrations/ # SQL migrations
│
├── frontend/ # React frontend
│ ├── src/
│ │ ├── components/ # UI components
│ │ ├── pages/ # Page views
│ │ ├── services/ # API clients
│ │ └── hooks/ # Custom React hooks
│ └── public/
│
├── app/ # Python backend (Legacy)
├── Docs/ # Documentation
└── .claude/ # Claude Code agents & tools

---

API Reference

Authentication

Endpoint	Method	Description
/api/auth/register	POST	Register new user
/api/auth/login	POST	User login
/api/auth/refresh	POST	Refresh access token
/api/auth/me	GET	Get current user profile

Documents

Endpoint	Method	Description
/api/documents	POST	Create document
/api/documents	GET	List documents (pagination, search)
/api/documents/:id	GET	Get document with AI analysis
/api/documents/:id	PUT	Update document
/api/documents/:id	DELETE	Delete document

Knowledge Search (RAG)

Endpoint	Method	Description
/api/search/ask	POST	Natural language Q&A
/api/search/semantic	POST	Vector similarity search
/api/search/korean	POST	Korean text search

Embeddings

Endpoint	Method	Description
/api/embeddings	POST	Generate embedding
/api/embeddings/document/:id	POST	Embed document
/api/embeddings/search	POST	Search by embedding

AI Analysis

Endpoint	Method	Description
/api/understanding/:id	GET	Get document understanding
/api/understanding/:id/analyze	POST	Analyze document
/api/ai/suggestions	POST	Get AI suggestions

Additional Endpoints

Tags & Categories

Endpoint	Method	Description
/api/tags	GET	List all tags
/api/tags	POST	Create tag
/api/tags/:slug	GET/PUT/DELETE	Tag CRUD
/api/categories	GET/POST	Category management

Comments & Ratings

Endpoint	Method	Description
/api/documents/:id/comments	GET/POST	Document comments
/api/comments/:id	PUT/DELETE	Comment CRUD
/api/documents/:id/rating	GET/POST/DELETE	Document ratings

Workflows

Endpoint	Method	Description
/api/documents/:id/workflow	GET	Get workflow info
/api/documents/:id/workflow/action	POST	Perform workflow action
/api/workflows/pending	GET	Pending reviews
/api/workflow-templates	GET/POST	Workflow templates

Admin & Security

Endpoint	Method	Description
/api/admin/users	GET	List users (admin)
/api/admin/stats	GET	System statistics
/api/security/status	GET	Security status
/api/security/logs	GET	Security logs

Full API documentation: Docs/API_DOCUMENTATION.md

---

Environment Variables

Required

Variable	Description
DATABASE_URL	PostgreSQL connection string
JWT_SECRET	JWT signing key (min 32 chars)

Optional

Variable	Default	Description
HOST	127.0.0.1	Server bind address
PORT	8000	Server port
DATABASE_MAX_CONNECTIONS	10	Connection pool size
JWT_EXPIRATION_HOURS	24	Token expiration
OPENSEARCH_URL	-	OpenSearch server
OPENAI_API_KEY	-	For embeddings
ANTHROPIC_API_KEY	-	For AI analysis
RUST_LOG	info	Logging level

---

Documentation

Document	Description
GETTING_STARTED.md	Detailed setup guide
ARCHITECTURE.md	System architecture deep-dive
API_DOCUMENTATION.md	API reference
SECURITY.md	Security guidelines
CLAUDE.md	Project context for AI assistants

---

Development

Running Tests

# Backend testscd minky-rust
cargo test# Frontend testscd frontend
npm test

Code Quality

# Rust linting
cargo clippy
# Rust formatting
cargo fmt
# Frontend lintingcd frontend
npm run lint

Development Commands (Claude Code)

This project includes Claude Code agents for development automation:

/pm # Start PM agent for task management
/next # Get next priority task
/ci start # Start CI/CD session
/health # Check system health
/review # Request code review

---

Contributing

We welcome contributions! Please see our contribution guidelines:

Development Workflow

1. Fork the repository
2. Create a feature branch (git checkout -b feat/amazing-feature)
3. Commit changes (git commit -m 'feat: Add amazing feature')
4. Push to branch (git push origin feat/amazing-feature)
5. Open a Pull Request

Commit Convention

feat: New feature
fix: Bug fix
refactor: Code refactoring
docs: Documentation
test: Tests
chore: Maintenance

Code Standards

• Rust: Follow cargo clippy recommendations
• React: ESLint + Prettier formatting
• Tests: Aim for 80% coverage
• Documentation: Update docs with code changes

---

Roadmap

• Phase 0: Rust backend migration, Agent system
• Phase 1: Document Understanding (Current)
  • pgvector integration
  • Document analysis pipeline
  • RAG search API
• Phase 2: Conversational Search
  • Chat interface
  • Streaming responses
• Phase 3: Knowledge Connections
  • Knowledge graph visualization
  • Auto-linking
• Phase 4: Tacit Knowledge Capture
  • Slack integration
  • Conversation mining

---

Legacy Python Backend

The original Python/Flask backend is available in the app/ directory for reference. It includes:

• Flask API with 140+ endpoints
• SQLAlchemy ORM
• JWT authentication
• Korean NLP support
• OpenSearch integration

The Rust backend (minky-rust/) is the active development target with improved performance and type safety.

---

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Acknowledgments

• Anthropic Claude - AI document understanding
• pgvector - Vector similarity search
• Axum - Rust web framework

---

Built with care for teams who value their collective knowledge.