sql-safety

Model Database Protocol — intent-based, secure database access for LLMs with schema validation, policy engine, data masking, and MCP server support

In-memory recursive SQL query validator checking syntax safety and blocking dangerous DDL/DML injection commands

In-memory recursive SQL query validator checking syntax safety and blocking dangerous DDL/DML injection commands

基于 RAG 检索增强 + LangGraph Agent + 多层安全校验的电影数据智能问答系统

安全闸门优先的中文自然语言查库：白名单 SQL 校验 + 只读沙箱

Natural language to guarded, read-only SQL: AST safety checks, a self-correcting agent loop, and an execution-accuracy eval. Runs offline or with any OpenAI-compatible LLM.

Text2SQL data-analysis agent: three handwritten LangGraph orchestrations (deterministic pipeline / autonomous ReAct loop / Supervisor multi-agent with LLM-as-Judge), 4-layer SQL safety, MCP server+client, human-in-the-loop write path with transactional dry-run. 60-case gold benchmark.

Claude Code skill: run read-only SQL on production Postgres safely, with a saved query library and a 3-layer SQL gate.

Text-to-SQL analytics copilot as a graph workflow (Google ADK 2.0): sqlglot AST validation (read-only, allow-listed, LIMIT-bounded) between the model and the database, sandboxed DuckDB execution, self-correction, charts, full-stack FastAPI + Next.js. Runs offline with an execution-accuracy gate.

Natural-language-to-SQL with RAG-grounded schema retrieval and multi-layer SQL safety validation (allow-list, anti-stacking, injection heuristics, LLM arbiter), caching, and per-datasource concurrency control

MCP server that gives AI agents a compact map of a database schema, finds join paths between tables, and validates SQL through an AST-based read-only guard (sqlglot).

自然语言驱动的企业数据分析 Agent。一句话查全公司数据,业务人员无需懂 SQL 即可查询企业数据库、数仓与 Excel。 支持 OpenAI / DeepSeek / Qwen / 豆包 / 智谱等多 LLM 切换;基于 BGE-M3 Embedding + 元数据过滤实现 Schema 语义检索; 内置 SQL 安全沙箱(关键字白名单 + 危险模式拦截 + 强制 LIMIT);ReAct 自检循环让 SQL 执行失败自动修复; 自动图表推荐与中文洞察摘要;预置电商 / SaaS / 教育 3 大行业 52 个核心指标模板,开箱即用。