PocketMCP
PocketMCP is a lightweight, local-first MCP (Model Context Protocol) server that automatically watches folders, chunks and embeds files locally using Transformers.js with MiniLM, stores vectors in SQLite + sqlite-vec, and exposes semantic search capabilities to VS Code and Cursor. Designed for small machines (I'm running on an Intel N100 with 16GB RAM) with zero external dependencies after initial model download.
Features
- 🔍 Semantic Search: Find content by meaning, not just keywords
- 📁 Auto-Ingestion: Watches folders and automatically processes new/changed files
- 📄 Multi-Format Support: PDF, DOCX, Markdown, and plain text files
- ⚡ Local-First: Runs completely offline after initial model download
- 🗄️ SQLite Storage: Fast, reliable vector storage with sqlite-vec extension
- 🔧 MCP Integration: Native support for VS Code and Cursor via MCP protocol
- 🌐 Web Interface: Built-in web tester for validation and manual testing
- 💾 Efficient: Designed for resource-constrained environments
- 🔄 Real-time: Debounced file watching with smart concurrency limits
- 📊 Smart Segmentation: Page-aware PDF processing and section-aware DOCX handling
- 🛡️ Robust Error Handling: Graceful handling of encrypted, corrupted, or oversized files
Architecture
flowchart TD
subgraph "MCP Clients"
A[VS Code]
B[Cursor]
end
subgraph "Web Interface"
W1[React Frontend<br/>:5173]
W2[Express API<br/>:5174]
end
subgraph "PocketMCP Server"
C[MCP Server<br/>stdio transport]
D[File Watcher<br/>chokidar]
E[Text Chunker<br/>~1000 chars]
F[Embeddings<br/>Transformers.js<br/>MiniLM-L6-v2]
G[SQLite + sqlite-vec<br/>Vector Database]
end
subgraph "File System"
H[Watch Directory<br/>./kb/]
I[Data Directory<br/>./data/]
end
A -.->|MCP Tools| C
B -.->|MCP Tools| C
W1 -->|HTTP API| W2
W2 -->|Database Access| G
C --> D
D -->|File Changes| E
E -->|Text Chunks| F
F -->|384-dim Vectors| G
G -.->|Search Results| C
D -.->|Monitors| H
G -.->|Stores in| I
classDef mcpClient fill:#e1f5fe
classDef webInterface fill:#fff3e0
classDef server fill:#f3e5f5
classDef storage fill:#e8f5e8
class A,B mcpClient
class W1,W2 webInterface
class C,D,E,F,G server
class H,I storage
Performance & Limits
- Sweet spot: 10K-100K chunks on modest hardware
- Query latency: Sub-100ms for
top_k <= 10on typical corpora - Memory usage: ~100MB for model + minimal overhead per document
- Concurrency: Limited to 3 simultaneous file operations by default
- File size limit: 50MB per file (configurable)
Screenshots
Table of Contents
- Features
- Architecture
- Performance & Limits
- Quick Start
- Web Tester
- MCP Client Integration
- API Reference
- Configuration
- Development
- Deployment
- Troubleshooting
Quick Start
1. Installation
# Clone or download the project
cd PocketMCP
# Install dependencies
pnpm install
# Setup environment
pnpm setup
# Or manually: cp .env.sample .env
2. Configuration
Edit .env file:
# SQLite database path
SQLITE_PATH=./data/index.db
# Directory to watch for file changes (optional)
WATCH_DIR=./kb
# Embedding model (default is recommended)
MODEL_ID=Xenova/all-MiniLM-L6-v2
# Chunking configuration
CHUNK_SIZE=1000
CHUNK_OVERLAP=120
3. Create Content Directory
# Create directory for your documents
mkdir -p kb
# Add some markdown or text files
echo "# My First Document" > kb/test.md
echo "This is a sample document for testing PocketMCP." >> kb/test.md
4. Start the Server
# Development - MCP server + web interface
pnpm dev
# Production - MCP server only
pnpm build && pnpm start
On first run, the server will download the MiniLM model (~100MB) and then process any files in your watch directory.
Web Tester
PocketMCP includes a comprehensive web interface for testing and validation.
Access Points
- Web Interface: http://127.0.0.1:5173
- API Server: http://127.0.0.1:5174
- Health Check: http://127.0.0.1:5174/health
Features
📊 Database Diagnostics Panel
- Real-time database status monitoring
- Table counts and vector dimensions
- SQLite WAL mode verification
- Error detection and reporting
- One-click smoke testing
🔍 Search Panel
- Interactive semantic search testing
- LIKE vs Vector search modes
- Configurable result count (top-K)
- Detailed result inspection
- Performance metrics (response time)
📄 Documents Panel
- Browse all indexed documents
- Pagination support
- Document metadata display
- Creation and update timestamps
🔎 Chunk Viewer
- Detailed chunk inspection modal
- Full text content display
- Metadata and offset information
- Copy-to-clipboard functionality
API Endpoints
| Endpoint | Method | Description |
|---|---|---|
/health | GET | Server health check |
/api/db/diag | GET | Database diagnostics |
/api/search | POST | Semantic search |
/api/chunk/:id | GET | Get specific chunk |
/api/docs | GET | List documents |
Example API Usage
Search Documents:
curl -X POST http://127.0.0.1:5174/api/search \
-H "Content-Type: application/json" \
-d '{"query": "machine learning", "top_k": 5, "mode": "like"}'
Get Diagnostics:
curl http://127.0.0.1:5174/api/db/diag | jq .
🔧 MCP Client Integration
Cursor Integration
- Open Cursor Settings → MCP
- Add a new server:
{
"command": "pnpm",
"args": ["dev:mcp"],
"cwd": "/path/to/PocketMCP",
"env": {
"TRANSPORT": "stdio",
"SQLITE_PATH": "./data/index.db",
"WATCH_DIR": "./kb"
}
}
VS Code Integration
Add to your MCP settings:
{
"mcpServers": {
"pocketmcp": {
"command": "pnpm",
"args": ["dev:mcp"],
"cwd": "/path/to/PocketMCP",
"env": {
"TRANSPORT": "stdio",
"SQLITE_PATH": "./data/index.db",
"WATCH_DIR": "./kb"
}
}
}
}
HTTP Transport (Web Clients)
For web clients or remote access:
{
"mcpServers": {
"pocketmcp": {
"transport": "http",
"url": "http://localhost:8001/mcp"
}
}
}
📚 API Reference
MCP Tools
search
Search for similar content using semantic search.
{
"query": "machine learning algorithms",
"top_k": 5,
"filter": {
"doc_ids": ["doc_123", "doc_456"]
}
}
upsert_documents
Insert or update documents programmatically.
{
"docs": [
{
"text": "Your document content here...",
"external_id": "my_doc_1",
"title": "Important Notes",
"metadata": {}
}
]
}
delete_documents
Delete documents by ID.
{
"doc_ids": ["doc_123"],
"external_ids": ["my_doc_1"]
}
list_documents
List all documents with pagination.
{
"page": {
"limit": 20
}
}
MCP Resources
PocketMCP provides resource URIs for accessing specific chunks:
- Format:
mcp+doc://<doc_id>#<chunk_id> - Returns: Complete chunk data including text, offsets, and metadata
Configuration
Environment Variables
| Variable | Default | Description |
|---|---|---|
SQLITE_PATH | ./data/index.db | Path to SQLite database file |
WATCH_DIR | (none) | Directory to watch for file changes |
MODEL_ID | Xenova/all-MiniLM-L6-v2 | Hugging Face model for embeddings |
CHUNK_SIZE | 1000 | Target chunk size in characters |
CHUNK_OVERLAP | 120 | Overlap between chunks in characters |
PDF_MAX_PAGES | 300 | Maximum pages to process in PDF files |
PDF_MIN_TEXT_CHARS | 500 | Minimum text characters required in PDFs |
DOC_MAX_BYTES | 10000000 | Maximum file size for DOCX files (10MB) |
DOCX_SPLIT_ON_HEADINGS | false | Split DOCX documents on headings (h1/h2) |
NODE_ENV | development | Environment mode |
VERBOSE_LOGGING | false | Enable detailed logs |
DEBUG_DOTENV | false | Enable dotenv debug output |
API_PORT | 5174 | Web API server port |
API_BIND | 127.0.0.1 | API server bind address |
TRANSPORT | both | MCP transport mode (stdio/http/both) |
HTTP_HOST | 0.0.0.0 | HTTP server bind address |
HTTP_PORT | 8001 | MCP server port |
LOG_LEVEL | info | Logging level (debug/info/warn/error) |
Available Scripts
| Script | Description |
|---|---|
pnpm dev | Start web interface + API server for testing |
pnpm dev:mcp | Start MCP server (both transports + file watching) |
pnpm build | Build all components |
pnpm start | Start production MCP server (both transports + file watching) |
pnpm setup | Create .env from template |
pnpm clean | Clean build artifacts and database |
Watch Directory
WATCH_DIRis optional - if not set, only manual document upserts work- Supported files:
.md,.txt,.pdf,.docx - File filtering: Automatically ignores temp files,
.DS_Store,node_modules, etc. - Nested directories: Recursively watches all subdirectories
Document Processing
Processing Pipeline: Documents → Segments → Chunks
- Documents: Top-level files with metadata
- Segments: Logical divisions (PDF pages, DOCX sections, etc.)
- Chunks: Text pieces optimized for embedding (~1000 chars)
Status Types: ok, skipped, needs_ocr, too_large, error
Supported File Types
- Markdown (
.md) - Plain text (
.txt) - PDF (
.pdf) - Text-based only, no OCR - DOCX (
.docx) - Microsoft Word documents
Notes:
- Encrypted/password-protected files are skipped
- Large files exceeding limits are marked as
too_large - Scanned PDFs requiring OCR are marked as
needs_ocr
Development
Project Structure
PocketMCP/ # Monorepo root
├── package.json # Workspace configuration
├── pnpm-workspace.yaml # pnpm workspace setup
├── .env # Environment variables
├── .env.sample # Environment template
├── apps/
│ ├── api/ # Express API server
│ │ ├── src/
│ │ │ ├── server.ts # Main API server
│ │ │ └── db.ts # Database manager
│ │ └── package.json
│ └── web/ # React + Vite frontend
│ ├── src/
│ │ ├── App.tsx # Main app component
│ │ ├── store.ts # Zustand state management
│ │ ├── api.ts # API client
│ │ └── components/ # UI components
│ └── package.json
├── src/ # Original MCP server
│ ├── server.ts # MCP server and main entry point
│ ├── db.ts # SQLite database with sqlite-vec
│ ├── embeddings.ts # Transformers.js embedding pipeline
│ ├── chunker.ts # Text chunking with sentence awareness
│ ├── ingest.ts # Generic document ingestion
│ ├── file-ingest.ts # File-specific ingestion logic
│ └── watcher.ts # File system watcher with debouncing
├── data/ # SQLite database storage
├── kb/ # Default watch directory (configurable)
└── README.md
Development Commands
# Install and setup
pnpm install
pnpm setup
# Development
pnpm dev # Web interface + API
pnpm dev:mcp # MCP server only
# Production
pnpm build
pnpm start
# Testing
curl http://127.0.0.1:5174/health
Deployment
Docker (Recommended)
Quick Start:
# Pull and run with all services (MCP + API + Web UI)
docker run -d \
--name pocketmcp \
--restart unless-stopped \
-p 8001:8001 \
-p 5174:5174 \
-p 5173:5173 \
-v pocketmcp_data:/app/data \
-v pocketmcp_kb:/app/kb \
-v pocketmcp_cache:/app/.cache \
ghcr.io/kailash-sankar/pocketmcp:latest
Access Points:
- MCP Server:
http://localhost:8001 - API Server:
http://localhost:5174 - Web UI:
http://localhost:5173
Docker Compose:
git clone https://github.com/kailash-sankar/PocketMCP.git
cd PocketMCP
cp .env.sample .env
docker-compose up -d
Portainer Stacks
- Go to Stacks → Add stack
- Name:
pocketmcp - Paste this configuration:
version: '3.8'
services:
pocketmcp:
image: ghcr.io/kailash-sankar/pocketmcp:latest
container_name: pocketmcp
restart: unless-stopped
ports:
- "8001:8001" # MCP Server
- "5174:5174" # API Server
- "5173:5173" # Web UI
volumes:
- pocketmcp_data:/app/data
- pocketmcp_kb:/app/kb
- pocketmcp_cache:/app/.cache
environment:
- NODE_ENV=production
- TRANSPORT=both
- SQLITE_PATH=/app/data/index.db
- WATCH_DIR=/app/kb
healthcheck:
test: ["CMD", "curl", "-f", "http://localhost:5173/health"]
interval: 30s
timeout: 10s
retries: 3
start_period: 60s
volumes:
pocketmcp_data:
pocketmcp_kb:
pocketmcp_cache:
Direct Installation
# Clone and setup
git clone https://github.com/kailash-sankar/PocketMCP.git
cd PocketMCP
pnpm install
pnpm setup
# Configure environment
cp .env.sample .env
# Edit .env with your settings
# Create content directory
mkdir -p kb
# Build and start
pnpm build
pnpm start
Access Points:
- MCP Server:
http://localhost:8001 - API Server:
http://localhost:5174 - Web UI:
http://localhost:5173
Troubleshooting
Model Download Issues
If the embedding model fails to download:
- Check internet connection for initial download
- Model cache location:
~/.cache/huggingface/transformers/ - Clear cache and retry if needed
SQLite Extension Issues
If sqlite-vec fails to load:
- Ensure
sqlite-vecnpm package is installed - Check that your system supports the required SQLite version
- The system automatically falls back to regular SQLite tables if vec0 virtual tables fail
File Watching Issues
- Files not being detected: Check file extensions and ignore patterns
- High CPU usage: Increase debounce time with larger
debounceMsvalues - Permission errors: Ensure read/write access to watch and data directories
Web Interface Issues
- API not accessible: Ensure API server is running on port 5174
- Database not found: Check
SQLITE_PATHenvironment variable - CORS errors: API server includes CORS headers for local development
Memory Issues
- Reduce
CHUNK_SIZEfor lower memory usage - Process fewer files simultaneously by reducing
maxConcurrency - Consider using a smaller embedding model (though this requires code changes)
Common Error Messages
"Too many parameter values were provided"
- This was a known issue with sqlite-vec virtual tables, now fixed with automatic fallback
"Failed to load sqlite-vec extension"
- System automatically falls back to regular SQLite tables with JSON embeddings
"Database file does not exist"
- Run the MCP server first to create the database, or check the
SQLITE_PATH
License
MIT License - see LICENSE file for details.
Contributing
- Fork the repository
- Create a feature branch
- Make your changes
- Add tests if applicable
- Submit a pull request
Acknowledgments
- sqlite-vec for fast vector similarity search
- Transformers.js for local embedding generation
- Model Context Protocol for standardized tool integration
- Hugging Face for the MiniLM model
- React + Vite for the modern web interface
- TailwindCSS for beautiful, responsive styling