TaskWeaver is A code-first agent framework for seamlessly planning and executing data analytics tasks. This innovative framework interprets user requests through code snippets and efficiently coordinates a variety of plugins in the form of functions to execute data analytics tasks in a stateful manner.
Unlike many agent frameworks that only track the chat history with LLMs in text, TaskWeaver preserves both the chat history and the code execution history, including the in-memory data. This feature enhances the expressiveness of the agent framework, making it ideal for processing complex data structures like high-dimensional tabular data.
- 📅2025-03-13: TaskWeaver now supports vision input for the Planner role. Please check the vision input for more details.👀
- 📅2025-01-16: TaskWeaver has been enhanced with an experimental role called Recepta for its reasoning power.🧠
- 📅2024-12-23: TaskWeaver has been integrated with the AgentOps for better observability and monitoring.🔍
- 📅2024-09-13: We introduce the shared memory to store information that is shared between the roles in TaskWeaver. Please check the memory for more details.🧠
- 📅2024-09-13: We have enhanced the experience feature by allowing static and dynamic experience selection. Please check the experience for more details.📚
- 📅2024-07-02: We have optimized TaskWeaver to support not-that-large language models served locally. Please check this post for more details.🔗
- 📅2024-05-07: We have added two blog posts on Evaluating a LLM agent and Adding new roles to TaskWeaver in the documentation.📝
- 📅2024-03-28: TaskWeaver now offers all-in-one Docker image, providing a convenient one-stop experience for users. Please check the docker for more details.🐳
- 📅2024-03-27: TaskWeaver now switches to
containermode by default for code execution. Please check the code execution for more details.🐳
- ......
- 📅2023-11-30: TaskWeaver is released on GitHub🎈.
- Planning for complex tasks - TaskWeaver, which features task decomposition and progress tracking, is designed to solve complex tasks.
- Reflective execution - TaskWeaver supports reflective execution, which allows the agent to reflect on the execution process and make adjustments.
- Rich data structure - TaskWeaver allows you to work with rich data structures in Python, such as DataFrames, instead of dealing with strings.
- Customized algorithms - TaskWeaver allows you to encapsulate your own algorithms into plugins and orchestrate them.
- Incorporating domain-specific knowledge - TaskWeaver is designed to incorporate domain-specific knowledge easily to improve the reliability.
- Stateful execution - TaskWeaver is designed to support stateful execution of the generated code to ensure consistent and smooth user experience.
- Code verification - TaskWeaver is designed to verify the generated code before execution. It can detect potential issues in the generated code and provide suggestions to fix them.
- Easy to use - TaskWeaver is easy to use with sample plugins, examples and tutorials to help you get started. TaskWeaver offers an open-box experience, allowing users to run it immediately after installation.
- Easy to debug - TaskWeaver is easy to debug with detailed and transparent logs to help you understand the entire process, including LLM prompts, the code generation, and execution process.
- Security consideration - TaskWeaver supports a basic session management to keep different users' data separate. The code execution is separated into different processes to avoid mutal interference.
- Easy extension - TaskWeaver is easy to extend to accomplish more complex tasks with multiple agents as roles and plugins.
There are still many features and improvements can be made. But due to our limited resources, we are not able to implement all of them or the progress will be slow. We are looking forward to your contributions to make TaskWeaver better.
- Easy-to-use and maintainable UX/UI
- Support for prompt template management
- Better plugin experiences, such as displaying updates or stopping in the middle of running the plugin and user confirmation before running the plugin
- Async interaction with LLMs
- Support for remote code execution
TaskWeaver requires Python >= 3.10. It can be installed by running the following command:
# [optional to create conda environment]
# conda create -n taskweaver python=3.10
# conda activate taskweaver
# clone the repository
git clone https://github.com/microsoft/TaskWeaver.git
cd TaskWeaver
# install the requirements
pip install -r requirements.txtIf you want to install an earlier version of TaskWeaver, you may check the release page, find the tag (e.g., v0.0.1) and install it by
pip install git+https://github.com/microsoft/TaskWeaver@<TAG>
Before running TaskWeaver, you need to provide your LLM configurations. Taking OpenAI as an example, you can configure taskweaver_config.json file as follows.
{
"llm.api_key": "the api key",
"llm.model": "the model name, e.g., gpt-4"
}💡 TaskWeaver also supports other LLMs and advanced configurations, please check the documents for more details.
💡 TaskWeaver has switched to container mode by default for code execution, which means the code is run in a container.
You may need to install Docker and take care of the dependencies in the container.
Please check the code execution for more details.
# assume you are in the cloned TaskWeaver folder
python -m taskweaver -p ./project/This will start the TaskWeaver process and you can interact with it through the command line interface. If everything goes well, you will see the following prompt:
=========================================================
_____ _ _ __
|_ _|_ _ ___| | _ | | / /__ ____ __ _____ _____
| |/ _` / __| |/ /| | /| / / _ \/ __ `/ | / / _ \/ ___/
| | (_| \__ \ < | |/ |/ / __/ /_/ /| |/ / __/ /
|_|\__,_|___/_|\_\|__/|__/\___/\__,_/ |___/\___/_/
=========================================================
TaskWeaver: I am TaskWeaver, an AI assistant. To get started, could you please enter your request?
Human: ___
TaskWeaver also supports WebUI for demo purpose, please refer to web UI docs for more details.
TaskWeaver can be imported as a library to integrate with your existing project, more information can be found in docs
More documentations can be found on TaskWeaver Website.
- ❔GitHub Issues (Preferred)
- 💬 Discord for discussion
- For other communications, please contact taskweaver@microsoft.com
The demos were made based on the web UI, which is better for displaying the generated artifacts such as images. The demos could also be conducted in the command line interface.
In this example, we will show you how to use TaskWeaver to pull data from a database and apply an anomaly detection algorithm.
anomaly_detection_chainlit.mp4
If you want to follow this example, you need to configure the sql_pull_data plugin in the project/plugins/sql_pull_data.yaml file.
You need to provide the following information:
api_type: azure or openai
api_base: ...
api_key: ...
api_version: ...
deployment_name: ...
sqlite_db_path: sqlite:///../../../sample_data/anomaly_detection.dbThe sql_pull_data plugin is a plugin that pulls data from a database. It takes a natural language request as input and returns a DataFrame as output.
This plugin is implemented based on Langchain. If you want to follow this example, you need to install the Langchain package:
pip install langchain
pip install tabulateIn this example, we will show you how to use TaskWeaver to forecast QQQ's price in the next 7 days.
forecast_price_chainlit.mp4
If you want to follow this example, you need to ensure you have these two requirements installed:
pip install yfinance
pip install statsmodelsFor more examples, please refer to our paper.
💡 The planning of TaskWeaver are based on the LLM model. Therefore, if you want to repeat the examples, the execution process may be different from what you see in the videos. For example, in the second demo, the assistant may ask the user which prediction algorithm should be used. Typically, more concrete prompts will help the model to generate better plans and code.
Our paper could be found here. If you use TaskWeaver in your research, please cite our paper:
@article{taskweaver,
title={TaskWeaver: A Code-First Agent Framework},
author={Bo Qiao, Liqun Li, Xu Zhang, Shilin He, Yu Kang, Chaoyun Zhang, Fangkai Yang, Hang Dong, Jue Zhang, Lu Wang, Minghua Ma, Pu Zhao, Si Qin, Xiaoting Qin, Chao Du, Yong Xu, Qingwei Lin, Saravan Rajmohan, Dongmei Zhang},
journal={arXiv preprint arXiv:2311.17541},
year={2023}
}
This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow Microsoft's Trademark & Brand Guidelines. Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party's policies.
The recommended models in this Repo are just examples, used to explore the potential of agent systems with the paper at TaskWeaver: A Code-First Agent Framework. Users can replace the models in this Repo according to their needs. When using the recommended models in this Repo, you need to comply with the licenses of these models respectively. Microsoft shall not be held liable for any infringement of third-party rights resulting from your usage of this repo. Users agree to defend, indemnify and hold Microsoft harmless from and against all damages, costs, and attorneys' fees in connection with any claims arising from this Repo. If anyone believes that this Repo infringes on your rights, please notify the project owner email.