queryclaw 0.1.0

AI-Native Database Agent for Safe & Autonomous Operations — Text-to-SQL with safety guardrails

🦞 QueryClaw

The AI-Native Database Agent for Safe & Autonomous Operations.

🚀 Mission: To bridge the gap between Natural Language and Transactional Database Operations with enterprise-grade safety.

QueryClaw is not just a Text-to-SQL chatbot. It is an intelligent agent that plans, validates, executes, and self-corrects database operations (CRUD + DDL) with built-in safety guardrails, transaction management, and audit trails.

✨ Key Features

• 🛡️ Safety First: AST-based static analysis, execution plan dry-runs, and auto-rollback mechanisms.
• 🧠 Multi-Agent Planning: Decomposes complex natural language tasks into atomic, transactional SQL steps.
• 🔗 Business Context Aware: RAG-powered understanding of schema, foreign keys, and business metrics.
• 🔄 Self-Healing: Automatically detects errors, rewrites SQL, and retries.
• 📜 Full Audit: Complete lineage from natural language prompt to before/after data snapshots.

🏗 Architecture

• Core: Python
• Protocol: Model Context Protocol (MCP) Ready
• Support: MySQL, PostgreSQL, SQLite (More coming soon)

🚀 Roadmap

• Phase 1: MVP - Safe Read-Only Queries & Schema RAG
• Phase 2: Write Operations with Dry-Run & Rollback
• Phase 3: Multi-Agent Planning & Human-in-the-Loop
• Phase 4: MCP Server & Ecosystem Integrations

🤝 Contributing

We welcome contributions! Please read our Contributing Guidelines before submitting PRs.

📄 License

Distributed under the Apache 2.0 License. See LICENSE for more information.

---

Made with ❤️ by the QueryClaw Team