Sqlew

关于

Sqlew is a Model Context Protocol (MCP) server designed to overcome the challenge of AI agents lacking persistent memory and structured access to architectural decisions. By transforming traditional, scattered Markdown ADRs into a queryable, AI-native decision database, sqlew allows agents to efficiently retrieve relevant decisions, detect duplicates, track constraints, and automatically capture new architectural choices from Plan Mode. This leads to significant token savings, faster queries, and ensures consistent architectural understanding across sessions and multiple AI agents.

主要功能

• Structured records for architectural decisions with metadata, tags, and relationships
• Multi-database backend support including SQLite, PostgreSQL, and MySQL
• Three-tier similarity detection system to prevent duplicate or conflicting decisions
• 2 GitHub stars
• Automatic capture of architectural decisions from Claude Code's Plan Mode
• Efficient SQL-based querying for AI agents to retrieve only relevant decisions

使用案例

• Facilitate ADR-driven development for architecture evolution and pattern standardization with AI assistance
• Provide persistent context and knowledge sharing for multi-session AI projects and long-term maintenance
• Enable multi-agent coordination by sharing architectural understanding through a unified decision database