QUITE 是 AI Skill Hub 本期精选Agent工作流之一。综合评分 7.5 分,整体质量较高。我们推荐使用将其纳入你的 AI 工具库,帮助提升工作效率。
QUITE:一个基于LLM代理的查询重写系统,超越规则。
QUITE 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
QUITE:一个基于LLM代理的查询重写系统,超越规则。
QUITE 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
# 方式一:pip 安装(推荐)
pip install quite
# 方式二:虚拟环境安装(推荐生产环境)
python -m venv .venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install quite
# 方式三:从源码安装(获取最新功能)
git clone https://github.com/Yuyang-Song/QUITE
cd QUITE
pip install -e .
# 验证安装
python -c "import quite; print('安装成功')"
# 命令行使用
quite --help
# 基本用法
quite input_file -o output_file
# Python 代码中调用
import quite
# 示例
result = quite.process("input")
print(result)
# quite 配置文件示例(config.yml) app: name: "quite" debug: false log_level: "INFO" # 运行时指定配置文件 quite --config config.yml # 或通过环境变量配置 export QUITE_API_KEY="your-key" export QUITE_OUTPUT_DIR="./output"
<p align="center"> <img src="figures/overview.png" width="1000px"> </p>
QUITE (<u>QU</u>ery rewr<u>ITE</u>) is a training-free and feedback-aware system based on LLM agents that rewrites SQL queries into semantically equivalent forms with significantly better performance, covering a broader range of query patterns and rewrite strategies.
The figure above presents the query rewrite workflow:
chmod +x ./scripts/evaluation.sh bash ./scripts/evaluation.sh
**Option 2: No-Restart Mode (for restricted environments)**bash
The following instructions have been tested on Ubuntu 22.04 and PostgreSQL v14.13, Python 3.12.
Install dependencies:
pip install -r requirements.txt Note: For the Rewrite Middleware dependencies installation, we have simplified and integrated the installation process as much as possible into our workflow. If there still has any error related to this during runtime, you can check the official document to find help. We use SQLSolver SIGMOD'24 as the start point of our hybrid SQL Corrector, haystack to build our knowledge base.
Benchmarks:
We use TPC-H, DSB, Clacite and SQLStorm to evaluate our system's performance. You can find the construction methods in the following links: TPC-H, DSB, Calcite, SQLStorm.
Find the .env file in the config_file/ directory. Our default base LLM selection is as follows:
```bash
python run.py \ --input_path dataset/queries/tpch_test.json \ --output_dir output/my_results \ --schema_file dataset/schemas/tpch_schemas.sql \ --enable_rewriter \ --enable_recommender
REASONING_MODEL_API_KEY=[your_api_key_here] REASONING_MODEL=deepseek-r1 REASONING_MODEL_URL=[your_model_url_here]
DECISION_MODEL_API_KEY=[your_api_key_here] DECISION_MODEL=claude-3-7-sonnet-20250219 DECISION_MODEL_URL=[your_model_url_here]
ASSISTANT_MODEL_API_KEY=[your_api_key_here] ASSISTANT_MODEL=claude-3-7-sonnet-20250219 ASSISTANT_MODEL_URL=[your_model_url_here]
DB_HOST=[localhost] DB_PORT=[5432] DB_NAME=[your_database_name] DB_USER=[your_username] DB_PASSWORD=[your_password]
PROJECT_ROOT=/path/to/QUITE
Note: Make sure you have added the **PROJECT_ROOT** path in the `.env` file.
#### 1.2 Test Each Module
We independently test each component of QUITE to ensure that it would function correctly in the new execution environment. The default database is **TPC-H Benchmark** in our code [test_module.py](test_module.py) and [test_query](dataset/queries/test_sql.sql).
If you use your own database benchmarks, exchange the [test_query](dataset/queries/test_sql.sql) to yours and reconfirm the `.env` file.bash
chmod +x ./scripts/run_quite.sh bash ./scripts/run_quite.sh
**Option 2: Query Rewriting Only**bash
QUITE是一个有潜力的AI工作流系统,提供了基于LLM代理的查询重写功能,值得关注。
该工具未明确声明开源协议,商业使用前请联系原作者确认授权范围,避免侵权风险。
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
经综合评估,QUITE 在Agent工作流赛道中表现稳健,质量良好。如果你已有明确的使用需求,可以直接上手体验;如果还在评估阶段,建议对比同类工具后再做决策。
| 原始名称 | QUITE |
| 原始描述 | 开源AI工作流:QUITE: A query rewrite system beyond rules with LLM agents.。⭐20 · Python |
| Topics | workflowpython |
| GitHub | https://github.com/Yuyang-Song/QUITE |
| 语言 | Python |
收录时间:2026-06-11 · 更新时间:2026-06-11 · License:未公布 · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端