rdsai-cli 0.1.5

AI-powered CLI tool for Relation database management and analysis

RDSAI CLI

English | 中文

---

RDSAI CLI is a next-generation, AI-powered RDS CLI that transforms how you interact with the database. You describe your intent in natural language or SQL, and the AI agent performs hybrid processing of both: orchestrating diagnostic tools, analyzing execution plans, and executing queries — all without leaving your terminal. From performance troubleshooting to schema exploration, it handles the complexity so you can focus on what truly matters.

✨ Features

• AI Assistant for MySQL — Ask in natural language (English / 中文均可), get optimized SQL, diagnostics, and explanations
• Smart SQL Handling — Auto-detects SQL vs natural language, supports SQL completer, query history, etc.
• Instant SQL Result Explanation — Press Ctrl+E after any SQL query to get AI-powered explanations of results or errors, helping you understand query outcomes and troubleshoot issues quickly
• Multi-Model LLM Support — Work with multiple providers and models (Qwen, OpenAI, DeepSeek, Anthropic, Gemini, OpenAI-compatible) and switch via /model, with thinking mode support for transparent reasoning and decision-making processes
• Database Schema Analysis — Generate comprehensive database analysis reports with AI-powered schema review, index optimization suggestions, compliance checking against Alibaba Database Development Standards, and actionable recommendations
• Sysbench Performance Benchmarking — AI-powered performance testing with automated workflow (prepare → run → cleanup), comprehensive analysis reports including MySQL configuration analysis, InnoDB status analysis, bottleneck identification, and optimization recommendations
• MCP (Model Context Protocol) Integration — Extend capabilities by connecting to external MCP servers, including Alibaba Cloud RDS OpenAPI for cloud RDS instance management, monitoring, and operations
• Safety First — Read-only by default; DDL/DML requires confirmation (unless YOLO mode)
• YOLO Mode — One toggle to auto-approve all actions when you know what you're doing
• SSL/TLS Support — Full SSL configuration (CA, client cert, key, mode)

📦 Installation

Requirements

• Python 3.13+
• Network access to your RDS instance (currently only MySQL is supported)
• API access to at least one LLM provider (Qwen / OpenAI / DeepSeek / Anthropic / Gemini / OpenAI-compatible)
• sysbench (optional, for /benchmark command) — Install from sysbench GitHub

Install from PyPI

We recommend using uv as the Python package manager for faster installation and better dependency resolution. For more installation options, see uv installation guide.

# Using uv (recommended - provides isolated environment)
uv tool install --python 3.13 rdsai-cli

# Or using pip (may cause dependency conflicts with other packages)
# Recommended: Use virtual environment: python -m venv .venv && source .venv/bin/activate
pip install rdsai-cli

After installation, the rdsai command will be available globally. Note: Using pip install may cause dependency conflicts with other Python packages. We recommend using uv tool install or installing in a virtual environment.

Install from source (for development)

git clone https://github.com/aliyun/rdsai-cli.git
cd rdsai-cli

# Using uv
uv sync
uv sync --extra dev  # with dev dependencies

# Or using pip
pip install -e ".[dev]"

For development installations, use uv run rdsai or activate the virtual environment first.

🚀 Quick Start

1. Launch the CLI

# Start without connection (interactive mode)
rdsai

# Connect via command line arguments
rdsai --host localhost -u root -p secret -D mydb

# With SSL
rdsai --host db.example.com -u admin -p secret \
  --ssl-mode REQUIRED --ssl-ca /path/to/ca.pem

# Custom port
rdsai --host localhost -P 3307 -u root -p secret

You can start the CLI without any connection parameters and connect later using the interactive /connect command:

$ rdsai
> /connect
# Interactive form will prompt for Host, Port, Username, Password, Database

Connection options:

Option	Short	Description	Default
--host	-h	Database host
--user	-u	Username
--password	-p	Password
--port	-P	Port	3306
--database	-D	Default database
--yolo	-y	Auto-approve all actions	off
--ssl-mode		SSL mode
--ssl-ca		CA certificate path
--ssl-cert		Client certificate path
--ssl-key		Client key path

SSL modes: DISABLED, PREFERRED, REQUIRED, VERIFY_CA, VERIFY_IDENTITY

2. Configure LLM

Use the interactive wizard to configure your LLM provider:

mysql> /setup

The wizard will walk you through:

1. Select Platform — Qwen, OpenAI, DeepSeek, Anthropic, Gemini, or a generic OpenAI-compatible endpoint
2. Configure API — Base URL (if needed), API Key, Model Name
3. Save & Apply — Configuration is persisted and the shell is reloaded automatically

Configuration file:

• Path: ~/.rdsai-cli/config.json
• Contains: providers, models, language, default model settings

You can edit this JSON manually for advanced setups.

📖 Basic Usage

SQL Execution

Plain SQL is executed directly against MySQL, with results formatted via Rich:

mysql> SELECT COUNT(*) FROM users;
mysql> SHOW CREATE TABLE orders;
mysql> EXPLAIN SELECT * FROM users WHERE email = '[e-mail]';
mysql> SELECT * FROM users LIMIT 10\G   -- vertical format

Quick Explain with Ctrl+E: After executing any SQL query, press Ctrl+E to get an AI-powered explanation of the result or error. This helps you understand:

• What the query result means
• Why a query might be slow or return unexpected results
• The cause of SQL errors and how to fix them

The explain hint (💡 Ctrl+E: Explain result or 💡 Ctrl+E: Explain error) is automatically shown after each query execution.

Natural Language

Just type what you need; the agent will call tools, run DDL SQL (with confirmation), and explain results:

mysql> analyze index usage on users table
mysql> show me slow queries from the last hour
mysql> check for lock waits
mysql> design an orders table for e-commerce
mysql> why this query is slow: SELECT * FROM users WHERE name LIKE '%john%'
mysql> find tables without primary keys
mysql> show me the replication status

The shell automatically:

• Detects whether input is SQL or natural language
• Records query execution history
• Injects the last query result into the AI context when helpful

Meta Commands

Meta commands start with / and never hit MySQL directly.

Command	Alias	Description
/connect	/conn	Connect to MySQL database interactively
/disconnect	/disconn	Disconnect from current database
/help	/h, /?	Show help and current status
/exit	/quit	Exit CLI
/version		Show CLI version
/setup		Interactive LLM configuration wizard
/reload		Reload configuration
/clear	/reset	Clear AI context (start fresh)
/compact		Compact AI context to save tokens
/yolo		Toggle YOLO mode (auto-approve actions)
/history	/hist	Show SQL query execution history
/model	/models	Manage LLM models (list/use/delete/info)
/research		Generate comprehensive database schema analysis report
/benchmark		Run sysbench performance test with AI-powered analysis
/mcp		Manage MCP servers (list/connect/disconnect/enable/disable)

You can still run shell commands via the built-in shell mode when prefixed appropriately (see in-shell help).

📚 Documentation

Core Features

• Database Schema Analysis (/research) — Comprehensive schema analysis and compliance checking against Alibaba Database Development Standards
• Performance Benchmarking (/benchmark) — AI-powered sysbench testing with automated workflow and comprehensive analysis reports
• MCP Integration — Extend capabilities by connecting to external MCP servers for cloud RDS management and more

Usage Scenarios

• SQL Result Explanation with Ctrl+E — Get instant AI-powered explanations of query results and errors using the Ctrl+E shortcut
• Slow Query Analysis & Optimization — Identify and optimize slow queries using AI-powered analysis
• Lock Wait & Deadlock Troubleshooting — Diagnose and resolve lock contention issues
• Database Schema Analysis & Compliance Review — Comprehensive schema review with compliance checking
• Performance Benchmarking & Optimization — Run performance tests and get optimization recommendations
• Cloud RDS Management with MCP — Manage cloud RDS instances directly from the CLI

⚡ YOLO Mode

YOLO mode skips confirmation prompts for potentially destructive actions (DDL/DML).

# Enable at startup
rdsai --host localhost -u root -p secret --yolo

# Toggle at runtime
mysql> /yolo on
mysql> /yolo off

Use this only in non-production or when you fully trust the actions being taken.

🔒 Security Notes

1. Read-Only by Default — The AI runs in a conservative mode; DDL/DML require explicit confirmation unless YOLO is on.
2. Confirmation Required — Every write operation surfaces the exact SQL for review before execution.
3. Credential Storage — API keys and model settings are stored in ~/.rdsai-cli/config.json; protect that file with proper OS permissions.
4. Transaction Safety — The shell warns you about uncommitted transactions when you attempt to exit.

See GitHub Issues for detailed tracking.

🤝 Contributing

We welcome contributions of all kinds! Please see CONTRIBUTING.md for:

• Development setup
• Code style guidelines
• Pull request process
• Issue reporting

📜 License

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

---

Enjoy building and debugging RDS systems with an AI agent in your terminal 😁