Show HN: DynoTable – The DynamoDB GUI with real SQL and your AI agent 展示 HN：DynoTable——支持真实 SQL 和 AI 代理的 DynamoDB GUI

TL;DR

• DynoTable is a local-first desktop client for Amazon DynamoDB that enables real SQL queries, visual joins, and an AI agent powered by AWS Bedrock—all using the user’s own credentials without data leaving their machine.
• It introduces Smart Tables, allowing users to visually define relationships between DynamoDB tables and perform live, read-only joins without schema migration or ETL.
• The SQL Workbench supports full SELECT statements with GROUP BY, COUNT, SUM, AVG, HAVING, and multi-table joins, all executed locally on fetched data to minimize DynamoDB cost and latency.
• An integrated AI agent reads schemas, drafts writes as reviewable diffs, and operates entirely within the user’s AWS account via Bedrock, ensuring no third-party access to data.
• External agents like Claude Code, Cursor, or Codex can connect via MCP (Model Context Protocol) with scoped permissions and staged commits, enhancing developer workflow while maintaining security.

Why It Matters

This tool addresses critical pain points in working with DynamoDB: its lack of native SQL support, difficulty in querying across multiple tables, and limited AI integration without compromising data sovereignty. By bringing SQL-like capabilities and secure, local AI assistance directly to developers’ machines, DynoTable lowers the barrier for complex analytics and automation on NoSQL workloads while aligning with enterprise security and compliance needs.

Technical Details

• Local-First Architecture: All data processing—including SQL execution, joins, and AI inference—occurs client-side; only raw requests are sent to AWS services (DynamoDB, Bedrock) using the user’s IAM roles or SSO credentials.
• Smart Tables Visual Join Canvas: Users draw edges between tables based on PKs or GSIs; DynoTable materializes joined views as projections that recompute instantly upon edits, without altering source data.
• SQL Workbench with PartiQL Compatibility: Supports standard SQL aggregations and joins over DynamoDB data by fetching minimal required rows via optimal access patterns (Query > Scan), then computing results locally.
• AI Agent Integration via Bedrock: Leverages Amazon Bedrock models through the user’s AWS account; prompts, schema, and table content never leave the user’s environment, with write operations requiring explicit approval via diff review.
• MCP Agent Connectivity: Exposes a local MCP server enabling external LLM agents (e.g., Cursor, Claude Code) to interact with DynamoDB under user-defined scopes (read-only, staging, full), with all changes held in a staging area until committed manually.

Industry Insight

The rise of local-first, credential-preserving AI tools for cloud databases signals a shift toward developer-controlled, privacy-aware infrastructure—especially important as organizations adopt more generative AI workflows. DynoTable’s model of combining SQL abstraction over NoSQL stores with secure, extensible agent interfaces sets a precedent for how database clients should evolve: not just as viewers, but as intelligent, collaborative platforms that respect both performance constraints and data governance policies. This approach could become standard for other managed services where direct API access lacks analytical depth or safe AI augmentation.

TL;DR

• DynoTable是一款本地优先的DynamoDB桌面客户端，支持实时SQL查询、可视化编辑及基于用户自有AWS Bedrock密钥的AI代理。
• 核心技术创新在于通过本地计算实现跨表连接（Joins）与聚合操作，弥补了原生DynamoDB的功能缺失，同时确保数据不出本地边界。
• 该工具解决了NoSQL数据库在复杂分析场景下的痛点，为开发者提供了比AWS控制台更轻量且功能更强大的替代方案。
• 采用MCP协议允许外部AI Agent（如Claude Code）安全接入，实现了“代理提议、人工确认”的可控写入机制。
• 提供免费试用及永久免费的基础浏览模式，高级功能按座位计费且无第三方加价，强调成本透明性。

为什么值得看

对于使用AWS DynamoDB的工程师和数据分析师而言，DynoTable打破了NoSQL数据库无法进行复杂关联查询和聚合分析的常规限制，显著降低了数据处理门槛。其独特的本地化架构和基于自有凭证的AI集成，有效解决了云工具中常见的数据隐私泄露风险及合规顾虑，是提升云端数据库开发效率的重要辅助工具。

技术解析

• 本地优先架构：所有数据处理、SQL解析及AI推理均在用户本地机器完成，不经过DynoTable服务器，数据仅通过直连方式与AWS交互，确保了极高的安全性。
• Smart Tables可视化连接：允许用户在画布上手动定义表间关系，系统会自动选择最低成本的读取路径（PK或GSI）来执行实时连接，生成的视图为只读投影，不影响源表结构。
• SQL Workbench引擎：支持标准的SELECT语句、GROUP BY、COUNT及HAVING子句，利用PartiQL兼容层将SQL转换为底层的DynamoDB访问模式，并在执行前预览RCU估算成本。
• AI Agent集成：基于Amazon Bedrock构建，Agent直接读取用户的Schema并生成查询草稿，关键特性包括自动识别最优索引（Query优于Scan）以及所有变更必须经过用户审查确认后才能提交。
• MCP外部代理连接：内置本地MCP服务器，支持Claude Code、Cursor等主流IDE插件通过OAuth标准协议连接，实行严格的权限隔离（只读/ staging/全量），防止外部Agent直接修改生产数据。

行业启示

• NoSQL增强工具化趋势：随着企业级应用对数据分析要求的提高，单纯的NoSQL存储已难以满足需求，具备SQL转换、可视化和AI辅助能力的中间件将成为提升数据库可用性的标准配置。
• 数据主权与安全合规：在日益严格的GDPR等法规下，“数据不出域”成为选型关键，本地化运行的客户端工具相比SaaS平台更能赢得金融、医疗等敏感行业客户的信任。
• AI Agent工作流重构：从“直接写入”转向“提议 - 审查 - 提交”的模式，确立了人机协作的安全边界，这种设计范式有望被推广至其他基础设施管理工具中，以降低自动化操作的风险。