scorpion-db
Production fights back.
Every database client ships with the same lie: you are trusted. DELETE without WHERE goes through. UPDATE without LIMIT goes through. DROP TABLE in production goes through. Scorpion-db treats that as a defect. Destructive queries are blocked unless you declare explicit intent. Before and after hook pipelines let teams encode their own guardrails. ScorpionAI adds natural language to SQL, query explain, optimization, and anomaly detection on top. Plugin drivers for Postgres, MySQL, and SQLite at launch; MongoDB and Redis from the community. No other database client treats production as a hazard surface.
What scorpion-db can do
- Refuses bare DELETE, bare UPDATE, and DROP TABLE in production by default; override requires explicit intent flag
- Before/after query hook pipeline: teams encode custom safety nets, audit trails, and compliance gates as plugins
- ScorpionAI: natural language to SQL, query explain, query optimization, anomaly detection
- Plugin-based drivers: Postgres, MySQL, and SQLite ship at launch; MongoDB and Redis from the community
- Multi-surface: CLI, Tauri desktop app, and MCP server (so AI agents query safely too)
- Per-environment safety profiles: dev is permissive, staging warns, production blocks
Hook pipeline at the core. Every query passes through configurable before and after hooks before it reaches the driver. The pipeline is the product: safety is not a feature flag, it is the architecture.
