# DDLBuilder LLM Context

## Product URL
- [DDLBuilder](https://ddl.xingkaixin.me/)
  Entry point of the web product.

## Purpose
- This file helps an LLM understand DDLBuilder capabilities and route users to the correct docs quickly.
- Audience: users who want AI assistance for schema design and SQL workflow in DDLBuilder.

## Product Identity
- DDLBuilder is a web application for table schema design and SQL generation.
- It supports practical modeling, DDL/DCL output, AI-assisted workflows, and collaboration.

## What DDLBuilder Can Do
- Generate table DDL from form-based schema configuration.
- Generate privilege DCL for authorization scenarios.
- Support multiple database dialects: MySQL, PostgreSQL, PostgreSQL (Citus), SQL Server, Oracle, MariaDB, TiDB, Dameng, OceanBase (MySQL), OceanBase (Oracle), Kingbase, GBase, PolarDB, and GaussDB.
- Configure fields, indexes, constraints, and table-level options.
- Create table relationships by dragging between fields in the ER diagram, then confirm cardinality, optionality, referential actions, and indexes in the relationship wizard.
- Import SQL and parse schema-related statements back into editable config.
- Use AI-assisted table design to produce draft schemas from natural language.
- Use AI Modify to adjust the current table with reviewable field, index, and table-level changes.
- Use AI Index Advisor to analyze typical query SQL and add recommended indexes.
- Run AI review on generated DDL and apply selected suggestions.
- Explain selected SQL snippets for review and communication.
- Manage drafts and saved tables for iterative design.
- Sync drafts, saved tables, folders, and trash across devices when signed in, with real-time and incremental sync support.
- Share schema views with read-only links and continue via editable copy.
- Compare schema changes and generate alter/rollback-related outputs.
- Configure partitioning (supported dialects) and sharding (PostgreSQL Citus flow).

## Scope and Limits
- AI outputs are drafts and suggestions; human validation is required before production execution.
- Shared links are read-only by default; editing requires creating a copy.
- Feature visibility depends on selected database type (for example, partitioning and sharding tabs).
- This file is a capability overview, not a full operational manual.

## Official Documentation (English)
- [Documentation Home](https://ddl.xingkaixin.me/docs/en/)
  Start here for the complete English documentation index.
- [Quick Start](https://ddl.xingkaixin.me/docs/en/basic/getting-started)
  First-run workflow to build and copy your first DDL/DCL output.
- [Core Concepts](https://ddl.xingkaixin.me/docs/en/basic/core-concepts)
  Definitions of workspace, draft, saved table, share copy, DDL, and DCL.
- [Table and Field Configuration](https://ddl.xingkaixin.me/docs/en/basic/table-and-fields)
  How to model table metadata and field-level configuration.
- [Indexes, Privileges, and Misc](https://ddl.xingkaixin.me/docs/en/basic/index-auth-misc)
  How to configure indexes, use AI Index Advisor for query-based index recommendations, manage access control, and set table-level options.
- [DDL Output and Sharing](https://ddl.xingkaixin.me/docs/en/basic/ddl-and-share)
  How to copy SQL outputs and collaborate through share links.
- [Saved Tables and Draft box](https://ddl.xingkaixin.me/docs/en/basic/saved-tables)
  How to persist, reload, and organize schema versions.
- [Import and Parse SQL](https://ddl.xingkaixin.me/docs/en/advanced/import-and-parse)
  How to import existing SQL and convert it into editable configuration.
- [AI-Assisted Table Design Workflow](https://ddl.xingkaixin.me/docs/en/advanced/ai-workflow)
  How to generate schema drafts through AI conversation and refine existing tables through reviewable AI changes.
- [Review and Explain SQL](https://ddl.xingkaixin.me/docs/en/advanced/review-and-explain)
  How to use AI review and SQL explanation for quality checks.
- [Partitioning and Sharding Configuration](https://ddl.xingkaixin.me/docs/en/advanced/partition-and-sharding)
  How to use partition and sharding related features by database type.
- [Foreign Key Configuration and ER Diagram](https://ddl.xingkaixin.me/docs/en/advanced/foreign-key-and-er)
  How to create foreign keys manually or through the ER relationship wizard and validate relationship semantics.
- [Change Diff and Rollback](https://ddl.xingkaixin.me/docs/en/advanced/diff-and-rollback)
  How to inspect schema differences and rollback-related SQL outputs.
- [Errors and Failure Handling](https://ddl.xingkaixin.me/docs/en/faq/common-errors)
  Troubleshooting guide for common failures and recovery paths.
- [Feature Entry and Visibility](https://ddl.xingkaixin.me/docs/en/faq/feature-visibility)
  Why specific tabs or buttons may appear or be hidden.
- [Sharing and Collaboration](https://ddl.xingkaixin.me/docs/en/faq/sharing-and-collaboration)
  Collaboration expectations and behavior around share links.
- [Changelog](https://ddl.xingkaixin.me/docs/en/changelog/changelog)
  Release history and feature evolution reference.

## How an LLM Should Use This File
- Identify the user goal first: new table design, SQL import, AI review, SQL explanation, sharing, or troubleshooting.
- Route the user to the best matching doc link above and summarize actionable next steps.
- Keep guidance aligned with the selected database dialect and visible feature set.

## 中文说明

## 产品地址
- [DDLBuilder](https://ddl.xingkaixin.me/)
  Web 产品入口。

## 文件用途
- 这份文件用于帮助 LLM 快速理解 DDLBuilder 的能力边界，并把用户路由到合适的文档页面。
- 适用对象：希望在 DDLBuilder 中借助 AI 完成表结构设计和 SQL 工作流的用户。

## 产品定位
- DDLBuilder 是一个用于表结构设计与 SQL 生成的 Web 应用。
- 它支持实用建模、DDL/DCL 输出、AI 辅助工作流与协作分享。

## DDLBuilder 能做什么
- 通过表单化的结构配置生成建表 DDL。
- 针对授权场景生成权限 DCL。
- 支持多种数据库方言：MySQL、PostgreSQL、PostgreSQL (Citus)、SQL Server、Oracle、MariaDB、TiDB、Dameng、OceanBase (MySQL)、OceanBase (Oracle)、Kingbase、GBase、PolarDB 和 GaussDB。
- 配置字段、索引、约束以及表级选项。
- 在 ER 图中拖动字段连接点创建表关系，并通过关系向导确认基数、可选性、引用动作和索引。
- 导入 SQL，并将结构相关语句解析回可编辑配置。
- 通过 AI 辅助建表，用自然语言生成表结构草稿。
- 通过 AI 修改当前表，生成可审阅的字段、索引与表级变更。
- 通过 AI 索引顾问分析典型查询 SQL，并添加推荐索引。
- 对生成的 DDL 执行 AI 评审，并应用选中的建议。
- 对选中的 SQL 片段进行解释，便于评审和沟通。
- 管理草稿和已保存表，支持迭代式设计。
- 登录后支持草稿、已保存表、文件夹和回收站在多设备间同步，并支持实时同步与增量同步。
- 通过只读分享链接共享结构视图，并基于可编辑副本继续修改。
- 对比结构变更，并生成 alter / rollback 相关输出。
- 配置分区能力（支持的数据库方言）和分片能力（PostgreSQL Citus 流程）。

## 范围与限制
- AI 输出属于草稿和建议，投入生产执行前仍需人工校验。
- 分享链接默认是只读的；如需编辑，需要先创建副本。
- 某些功能是否可见取决于当前选择的数据库类型，例如分区和分片相关标签页。
- 这份文件提供的是能力概览，不是完整的操作手册。

## 官方文档（中文）
- [文档首页](https://ddl.xingkaixin.me/docs/zh/)
  从这里进入完整的中文文档目录。
- [快速开始](https://ddl.xingkaixin.me/docs/zh/basic/getting-started)
  首次使用流程，帮助你完成第一份 DDL / DCL 的配置与复制。
- [核心概念](https://ddl.xingkaixin.me/docs/zh/basic/core-concepts)
  了解工作区、草稿、已保存表、分享副本、DDL、DCL 等核心概念。
- [表与字段配置](https://ddl.xingkaixin.me/docs/zh/basic/table-and-fields)
  说明如何配置表元信息和字段级属性。
- [索引、权限与杂项](https://ddl.xingkaixin.me/docs/zh/basic/index-auth-misc)
  说明如何配置索引、使用 AI 索引顾问获取基于查询的索引建议、配置权限控制和表级选项。
- [DDL 输出与分享](https://ddl.xingkaixin.me/docs/zh/basic/ddl-and-share)
  说明如何复制 SQL 输出，以及如何通过分享链接协作。
- [已保存表与草稿箱](https://ddl.xingkaixin.me/docs/zh/basic/saved-tables)
  说明如何持久化、重新加载和整理不同版本的表结构。
- [导入与解析 SQL](https://ddl.xingkaixin.me/docs/zh/advanced/import-and-parse)
  说明如何导入已有 SQL，并转换为可编辑配置。
- [AI 辅助建表流程](https://ddl.xingkaixin.me/docs/zh/advanced/ai-workflow)
  说明如何通过 AI 对话生成表结构草稿，并通过可审阅的 AI 变更细化已有表。
- [评审与解释 SQL](https://ddl.xingkaixin.me/docs/zh/advanced/review-and-explain)
  说明如何使用 AI 评审和 SQL 解释能力做质量检查。
- [分区与分片配置](https://ddl.xingkaixin.me/docs/zh/advanced/partition-and-sharding)
  说明如何按数据库类型使用分区与分片相关能力。
- [外键配置与 ER 图](https://ddl.xingkaixin.me/docs/zh/advanced/foreign-key-and-er)
  说明如何手工配置外键，或通过 ER 关系向导创建关系并校验关系语义。
- [变更对比与回滚](https://ddl.xingkaixin.me/docs/zh/advanced/diff-and-rollback)
  说明如何查看结构差异以及回滚相关 SQL 输出。
- [常见错误与失败处理](https://ddl.xingkaixin.me/docs/zh/faq/common-errors)
  常见失败场景和恢复路径的排查说明。
- [功能入口与显示规则](https://ddl.xingkaixin.me/docs/zh/faq/feature-visibility)
  说明为什么某些标签页或按钮会显示或隐藏。
- [分享与协作](https://ddl.xingkaixin.me/docs/zh/faq/sharing-and-collaboration)
  说明分享链接场景下的协作方式与行为边界。
- [更新日志](https://ddl.xingkaixin.me/docs/zh/changelog/changelog)
  用于查看版本演进和功能变更记录。

## LLM 应如何使用这份文件
- 先识别用户目标：新建表设计、SQL 导入、AI 评审、SQL 解释、分享协作，还是问题排查。
- 从上面的文档链接中选择最匹配的一项，并给出可执行的下一步建议。
- 回答时要和当前选择的数据库方言以及界面中可见的功能保持一致。