Firestore MCP Server
# Firestore MCP Server
A Model Context Protocol (MCP) server that provides secure, permission-controlled access to Firebase Firestore. This server allows AI assistants and other MCP clients to interact with Firestore databases through a standardized interface.
## Features
### Core Functionality
- 🔐 **Granular Permissions**: Control access at the collection and operation level
- 📄 **Full CRUD Operations**: Create, read, update, and delete documents
- 🔍 **Advanced Queries**: Support for filtering, ordering, and limiting results
- 📁 **Subcollection Support**: Work with nested collections and documents
- 🔄 **Batch Operations**: Execute multiple operations atomically
- 💾 **Transactions**: Ensure data consistency with transactional operations
- 🎯 **Field Value Operations**: Atomic increments, array operations, and server timestamps
### Security & Control
- ✅ Collection-level access control
- 🛡️ Operation-specific permissions (read, write, delete, query)
- 🔒 Default deny with explicit allow rules
- 📋 Conditional permissions (coming soon)
## Installation
### Using npm/yarn/pnpm
```bash
npm install mcp-firestore
# or
yarn add mcp-firestore
# or
pnpm add mcp-firestore
```
### From Source
```bash
git clone https://github.com/yourusername/mcp-firestore.git
cd mcp-firestore
pnpm install
pnpm build
```
## Configuration
### Environment Variables
Create a `.env` file with your Firestore configuration:
```env
# Required
FIRESTORE_PROJECT_ID=your-project-id
# Optional - for authentication
GOOGLE_APPLICATION_CREDENTIALS=path/to/service-account.json
```
### Permission Configuration
Create a `permissions.json` file to control access:
```json
{
"collections": [
{
"collectionId": "users",
"operations": ["read", "write", "query"]
},
{
"collectionId": "posts",
"operations": ["read", "query"]
}
],
"defaultAllow": false
}
```
## Usage
### Starting the Server
```bash
# With default permissions
mcp-firestore
# With custom permissions file
mcp-firestore --config permissions.json
# With full access (development only)
mcp-firestore --full-access
# With read-only access to specific collections
mcp-firestore --read-only --collections users,posts
```
### Claude Desktop Integration
Add to your Claude Desktop configuration:
```json
{
"servers": {
"firestore": {
"command": "mcp-firestore",
"args": ["--config", "path/to/permissions.json"],
"env": {
"FIRESTORE_PROJECT_ID": "your-project-id"
}
}
}
}
```
## Available Tools
### Basic Operations
1. **firestore-list-collections**
- List all accessible collections
```json
{}
```
2. **firestore-get-collection**
- Get all documents from a collection
```json
{
"collectionId": "users"
}
```
3. **firestore-get-document**
- Get a specific document
```json
{
"collectionId": "users",
"documentId": "user123"
}
```
4. **firestore-create-document**
- Create a new document
```json
{
"collectionId": "users",
"documentId": "user123",
"data": {
"name": "John Doe",
"email": "john@example.com"
}
}
```
5. **firestore-update-document**
- Update an existing document
```json
{
"collectionId": "users",
"documentId": "user123",
"data": {
"name": "Jane Doe"
}
}
```
6. **firestore-delete-document**
- Delete a document
```json
{
"collectionId": "users",
"documentId": "user123"
}
```
### Query Operations
7. **firestore-query-collection**
- Query documents with filters
```json
{
"collectionId": "users",
"filters": [
{
"field": "age",
"operator": ">",
"value": 18
}
],
"orderBy": {
"field": "createdAt",
"direction": "desc"
},
"limit": 10
}
```
### Subcollection Operations
8. **firestore-list-subcollections**
- List subcollections of a document
```json
{
"documentPath": "users/user123"
}
```
9. **firestore-get-collection-by-path**
- Get documents from a subcollection
```json
{
"collectionPath": "users/user123/orders"
}
```
10. **firestore-create-document-by-path**
- Create a document in a subcollection
```json
{
"collectionPath": "users/user123/orders",
"data": {
"item": "Widget",
"quantity": 2
}
}
```
### Batch Operations
11. **firestore-batch-write**
- Execute multiple write operations atomically
```json
{
"operations": [
{
"type": "create",
"collectionPath": "products",
"documentId": "product1",
"data": { "name": "Widget" }
},
{
"type": "update",
"documentPath": "inventory/product1",
"data": { "count": 100 }
}
]
}
```
12. **firestore-batch-read**
- Read multiple documents in one operation
```json
{
"documentPaths": [
"users/user1",
"users/user2",
"products/product1"
]
}
```
13. **firestore-transaction**
- Execute a transaction with reads and conditional writes
```json
{
"reads": ["products/product1"],
"operations": [
{
"type": "update",
"documentPath": "products/product1",
"data": { "stock": 99 }
}
],
"conditionScript": "return readResults['products/product1'].data.stock > 0;"
}
```
### Field Value Operations
14. **firestore-increment-field**
- Atomically increment a numeric field
```json
{
"documentPath": "stats/daily",
"field": "visitCount",
"incrementBy": 1
}
```
15. **firestore-array-union**
- Add elements to an array without duplicates
```json
{
"documentPath": "users/user123",
"field": "tags",
"elements": ["premium", "verified"]
}
```
16. **firestore-server-timestamp**
- Set fields to server timestamp
```json
{
"documentPath": "users/user123",
"fields": ["lastLogin", "modifiedAt"]
}
```
## Resources
The server also provides MCP resources for direct access to Firestore data:
- `firestore://collections` - List all collections
- `firestore://collection/{collectionId}` - Access collection data
- `firestore://collection/{collectionId}/document/{documentId}` - Access document data
- `firestore://path/{path}` - Access any path (collections or documents)
## Examples
### Basic CRUD Operations
```javascript
// Create a new user
await client.callTool("firestore-create-document", {
collectionId: "users",
documentId: "user123",
data: {
name: "John Doe",
email: "john@example.com",
createdAt: new Date().toISOString()
}
});
// Update user data
await client.callTool("firestore-update-document", {
collectionId: "users",
documentId: "user123",
data: {
lastLogin: new Date().toISOString()
}
});
// Query active users
await client.callTool("firestore-query-collection", {
collectionId: "users",
filters: [
{ field: "status", operator: "==", value: "active" }
],
orderBy: { field: "createdAt", direction: "desc" },
limit: 10
});
```
### Working with Subcollections
```javascript
// Create an order for a user
await client.callTool("firestore-create-document-by-path", {
collectionPath: "users/user123/orders",
data: {
items: ["widget1", "widget2"],
total: 99.99,
status: "pending"
}
});
// Get all orders for a user
await client.callTool("firestore-get-collection-by-path", {
collectionPath: "users/user123/orders"
});
```
### Batch Operations
```javascript
// Atomic updates across multiple documents
await client.callTool("firestore-batch-write", {
operations: [
{
type: "update",
documentPath: "products/widget1",
data: { stock: 95 }
},
{
type: "create",
collectionPath: "orders",
data: {
product: "widget1",
quantity: 5,
userId: "user123"
}
},
{
type: "update",
documentPath: "users/user123",
data: { orderCount: 1 }
}
]
});
```
### Field Value Operations
```javascript
// Increment a counter
await client.callTool("firestore-increment-field", {
documentPath: "stats/global",
field: "totalOrders",
incrementBy: 1
});
// Add tags without duplicates
await client.callTool("firestore-array-union", {
documentPath: "products/widget1",
field: "tags",
elements: ["bestseller", "featured"]
});
// Set server timestamp
await client.callTool("firestore-server-timestamp", {
documentPath: "logs/access",
fields: ["timestamp", "lastModified"]
});
```
## Security Best Practices
1. **Use Minimal Permissions**: Only grant access to collections and operations that are necessary
2. **Default Deny**: Set `defaultAllow: false` in production environments
3. **Service Account Security**: Protect your service account credentials
4. **Environment Variables**: Never commit credentials to version control
5. **Audit Access**: Regularly review permission configurations
## Development
### Running Tests
```bash
pnpm test
pnpm test:coverage
```
### Building
```bash
pnpm build
```
### Docker
```bash
# Build image
docker build -t mcp-firestore .
# Run container
docker run -e FIRESTORE_PROJECT_ID=your-project mcp-firestore
```
## Troubleshooting
### Common Issues
1. **Authentication Errors**
- Ensure `GOOGLE_APPLICATION_CREDENTIALS` points to a valid service account
- Check that the service account has necessary Firestore permissions
2. **Permission Denied**
- Verify collection is listed in permissions configuration
- Check that the operation is allowed for the collection
3. **Connection Issues**
- Confirm project ID is correct
- Check network connectivity to Firestore
## Contributing
Contributions are welcome! Please read our [Contributing Guide](CONTRIBUTING.md) for details.
## License
MIT License - see [LICENSE](LICENSE) for details.
## Acknowledgments
Built on the [Model Context Protocol](https://modelcontextprotocol.io) by Anthropic.TDQS
Scored across 23 tools
Most tools have distinct purposes, but there is some overlap between 'firestore-get-collection' and 'firestore-query-collection' (both retrieve documents from a collection), and between 'firestore-create-document' and 'firestore-create-document-by-path' (both create documents, differing only in path specification). Descriptions clarify these differences, but an agent might occasionally misselect between the overlapping pairs.
All tool names follow a consistent 'firestore-verb-noun' pattern with hyphens, using clear verbs like 'create', 'get', 'update', 'delete', 'list', 'query', and 'increment'. The naming is highly predictable and readable throughout the set.
With 23 tools, the count is borderline high for a database server, potentially feeling heavy. However, it covers both basic CRUD operations and advanced Firestore-specific features like array updates, transactions, and batch operations, which justifies the number but may overwhelm agents.
The tool set provides comprehensive coverage for Firestore operations, including full CRUD lifecycle, batch and transaction support, field-specific updates (e.g., array operations, increments), and path-based variants for subcollections. No obvious gaps exist for typical agent workflows in this domain.