Mongosh - MongoDB Shell with SQL Support

A high-performance MongoDB shell written in Rust that bridges SQL and MongoDB - query your MongoDB databases using familiar SQL syntax or native MongoDB commands.

Note: This is an independent community project, not affiliated with MongoDB Inc.

🎯 Why Mongosh?

• 🔄 SQL to MongoDB - Write SQL queries, execute as MongoDB commands automatically
• ⚡ Blazing Fast - Native Rust implementation with async I/O
• 🎨 Beautiful Output - Syntax highlighting, formatted tables, and pretty JSON
• 🧠 Smart Completion - Context-aware auto-completion for collections, fields, and commands
• 💾 Named Queries - Save and reuse complex queries with parameters
• 📊 Rich SQL Features - Array indexing, date functions, arithmetic operations, and more
• 🤖 MCP Server - Let AI assistants query MongoDB with 15 built-in tools

📦 Installation

cargo install mongosh

🚀 Quick Start

# Connect to MongoDB
mongosh mongodb://localhost:27017

# Use SQL syntax
SELECT name, email FROM users WHERE age > 18 ORDER BY name LIMIT 10

# Or native MongoDB syntax
db.users.find({ age: { $gt: 18 } }).sort({ name: 1 }).limit(10)

# Save frequently used queries
query save active_users db.users.find({status: 'active'})
query active_users

✨ Key Features

1. SQL Query Support

Query MongoDB using standard SQL syntax - automatically translated to MongoDB queries:

-- Basic queries
SELECT * FROM orders WHERE status = 'completed'

-- Aggregations
SELECT category, COUNT(*) as total, AVG(price) as avg_price
FROM products
GROUP BY category
HAVING total > 10

-- Array access
SELECT tags[0] AS primary_tag FROM posts WHERE tags[-1] = 'featured'

-- Date filtering
SELECT * FROM events WHERE created_at > DATE '2024-01-01'

-- Arithmetic operations
SELECT price * quantity * 1.13 AS total FROM orders

-- Query analysis
EXPLAIN SELECT * FROM users WHERE age > 18

2. Named Queries with Parameters

Save and reuse queries with parameter substitution:

// Save a parameterized query
query save user_by_email db.users.findOne({email: '$1'})

// Execute with parameters
query user_by_email john@example.com

// List all saved queries
query list

3. Smart Shell Completion

Auto-complete datasource names from your configuration:

# Tab completion for datasources
mongosh -d prod<TAB>     # Completes to: mongosh -d production

# Works with options too
mongosh --datasource dev<TAB>

4. Rich Output Formats

# Pretty JSON (default)
db.users.find().limit(1)

# Compact JSON
mongosh --format json-compact

# Table format
mongosh --format table

# Shell-style output
mongosh --format shell

5. MCP Server (Model Context Protocol)

Enable AI assistants (Claude, Cursor, etc.) to query MongoDB directly:

mongosh --mcp mongodb://localhost:27017 --database mydb

Claude Desktop configuration (claude_desktop_config.json):

{
  "mcpServers": {
    "mongodb": {
      "command": "mongosh",
      "args": ["--mcp", "mongodb://localhost:27017", "--database", "mydb"]
    }
  }
}

15 built-in tools — full CRUD, aggregation, and admin operations:

Category	Tools
Context	mongo_prepare_context, mongo_list_datasources
Read	mongo_find, mongo_find_one, mongo_aggregate, mongo_count, mongo_distinct
Write	mongo_insert_one, mongo_insert_many, mongo_update_one, mongo_update_many
Delete	mongo_delete_one, mongo_delete_many
Admin	mongo_list_databases, mongo_list_collections, mongo_list_indexes, mongo_collection_stats, mongo_explain

Security — configure in ~/.mongoshrc:

[mcp.security]
allow_read = true
allow_write = false
allow_delete = false
max_documents_per_query = 1000
forbidden_collections = ["system.*", "admin.*"]

See MCP Documentation for details.

📚 Documentation

Comprehensive guides for all features:

• SQL Array Indexing - Access array elements with arr[0] or arr[-1]
• SQL Arithmetic Operations - Math expressions and functions
• DateTime Functions - DATE, TIMESTAMP, CURRENT_DATE
• Query EXPLAIN - Analyze query performance
• Named Queries - Save and reuse queries
• Shell Completion - Setup auto-completion
• API Reference - MongoDB method support status
• MCP Server - AI assistant integration via MCP

🆚 vs Official mongosh

Feature	Official mongosh	This Project
Language	JavaScript/Node.js	Rust
SQL Support	❌	✅ Full SQL SELECT
Startup Time	~500ms	<50ms
Memory Usage	~50MB	~5MB
Output Formats	JSON	JSON, Table, Shell
Named Queries	❌	✅ With parameters
Array Indexing	MongoDB only	SQL + MongoDB
Auto-completion	Basic	Advanced (datasources, fields)
MCP Server	❌	✅ 15 tools with security
JavaScript Runtime	✅ Full	❌ Not a JS shell

🔧 Configuration

Create ~/.mongoshrc or ~/.config/mongosh/config.toml:

[connection]
default_datasource = "local"

[connection.datasources]
local = "mongodb://localhost:27017"
production = "mongodb://prod.example.com:27017"
staging = "mongodb://staging.example.com:27017"

[output]
format = "pretty-json"
highlight = true

🤝 Contributing

Contributions are welcome! Please check out our documentation for implementation details.

📄 License

Licensed under the MIT License.

🔗 Links

• Documentation
• Crates.io
• GitHub Issues