Metadata-Version: 2.4
Name: co-gym
Version: 0.1.4
Summary: Framework and toolkits for building and evaluating collaborative agents that can work together with humans
Author-email: Yijia Shao <shaoyj@stanford.edu>, Vinay Samuel <vsamuel@umd.edu>
Maintainer-email: Yijia Shao <shaoyj@stanford.edu>, Vinay Samuel <vsamuel@umd.edu>
License-Expression: MIT
Project-URL: Homepage, https://cogym.saltlab.stanford.edu/
Project-URL: Repository, https://github.com/SALT-NLP/collaborative-gym
Project-URL: Documentation, https://cogym.saltlab.stanford.edu/
Project-URL: Bug Tracker, https://github.com/SALT-NLP/collaborative-gym/issues
Keywords: collaborative-ai,multi-agent,human-ai-collaboration,reinforcement-learning
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Developers
Classifier: Intended Audience :: Science/Research
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.8
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Requires-Python: >=3.11
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: autogen
Requires-Dist: ag2[jupyter-executor]
Requires-Dist: httpx<0.28.0,>=0.23.0
Requires-Dist: litellm>=1.59.3
Requires-Dist: knowledge-storm>=0.2.8
Requires-Dist: rstr
Requires-Dist: gymnasium
Requires-Dist: prompt-toolkit
Requires-Dist: websocket-client>=1.7.0
Requires-Dist: jupyter
Requires-Dist: aact>=0.0.10
Requires-Dist: typer
Requires-Dist: pydantic<3.0.0,>=2.8.0
Requires-Dist: fastapi
Requires-Dist: docker
Requires-Dist: pygments
Requires-Dist: datasets
Requires-Dist: scikit-learn
Requires-Dist: PyPDF2
Requires-Dist: uvicorn[standard]
Requires-Dist: arxiv
Requires-Dist: googlemaps
Requires-Dist: voyageai
Requires-Dist: python-multipart
Requires-Dist: anthropic
Requires-Dist: apscheduler
Provides-Extra: dev
Requires-Dist: pytest>=6.0; extra == "dev"
Requires-Dist: black>=22.0; extra == "dev"
Requires-Dist: flake8>=4.0; extra == "dev"
Requires-Dist: mypy>=0.950; extra == "dev"
Provides-Extra: docs
Requires-Dist: sphinx>=4.0; extra == "docs"
Requires-Dist: sphinx-rtd-theme>=1.0; extra == "docs"
Dynamic: license-file

<p align="center">
  <img src="assets/logo.svg" style="width: 60%; height: auto;">
</p>

---

<p align="center">
  <img src="assets/teaser-image-github.png" style="width: 100%; height: auto;">
</p>

**Latest News** 🔥

- [2025/02] 🎉 Co-Gym (real) is now publicly available! **Visit [our website](https://cogym.saltlab.stanford.edu/) to use collaborative agents for travel planning and tabular analysis.** We've also open-sourced our UI components for developers to build upon.
- [2025/01] Initial release with arXiv [preprint](https://arxiv.org/abs/2412.15701) (check out our [Twitter thread](https://x.com/EchoShao8899/status/1880291718496670097) for walkthrough video). We are working on releasing Co-Gym (real) web preview for in-the-wild study, alongside tools for developing collaborative agents locally.

## What is Collaborative Gym? [(Use Collaborative Agents Now!)](https://cogym.saltlab.stanford.edu/)
Collaborative Gym (Co-Gym) is a framework for enabling and evaluating **Human-Agent Collaboration**. Unlike traditional chatbots, Co-Gym allows both human users and AI agents to not only exchange messages but also take actions in a shared workspace. Additionally, unlike autonomous agents that take a sequence of actions on its own, Co-Gym expects the agents to engage in the collaboration process where the human user can intervene in real-time.

Our vision is to leverage Co-Gym for improving collaborative agents and evaluating human-agent collaboration.

- [Setup](#setup)
- [API](#api)
- [Simulated and Real Conditions](#simulated-and-real-conditions)
- [Evaluation](#evaluation)
- [Add a New Agent](#add-a-new-agent)
- [Add a New Task Environment](#add-a-new-task-environment)
- [Contributions](#contributions)
- [Acknowledgement](#acknowledgement)
- [Citation](#citation)


## Setup
**We view Co-Gym as a framework for building and evaluating collaborative agents that take initiative at the correct time and work together with humans in parallel.** Below, we provide a quick start guide to run Co-Gym locally to reproduce our experiments.

1. Install the required packages:
   ```shell
   conda create -n cogym python=3.11
   conda activate cogym
   pip install -r requirements.txt
   ```
2. Set up API keys: Copy `secrets.example.toml` from the root directory and rename it to `secrets.toml`. Complete the required fields by following the instructions provided in the comment.
3. To use the Jupyter Executor in some task environments, create the docker image for execution:
    ```shell
    cd docker
    docker build -f Dockerfile_cpu -t cogym-jupyter-cpu-image .
    ```

## API
The Co-Gym API models the environment where humans and agents can work together as Python [`CoEnv`](collaborative_gym/core.py) classes. We currently support three task environments:
- `CoTravelPlanningEnv(CoEnv)`: Collaborative environment for planning travel itineraries, supporting several search functionalities, distance matrix, and editing travel plans.
- `CoLitSurveyEnv(CoEnv)`: Collaborative environment for conducting literature survey (e.g., writing Related Work section for research), supporting searching for papers, taking notes, and writing/polishing a related works section.
- `CoAnalysisEnv(CoEnv)`: Collaborative environment for analyzing tabular data in csv format, supporting executing code in Jupyter notebook and documenting findings.

Creating environment instances and interacting with them is simple - here's an example using `CoTravelPlanningEnv`:
```python
from collaborative_gym.envs import EnvFactory
env = EnvFactory.make(
  name="travel_planning",
  team_members=["agent", "human"],
  env_id="env_1234",
  use_simulated_dataset=False,
  query="Help me plan a 5-day trip to Vancouver in December for a single person."
)

# In asynchronous collaboration, at a certain timestamp,
# agent can choose to take a task action like the following
obs, reward, terminated, private, info = env.step(
  role="agent",
  action="EDITOR_UPDATE(text=\"This is a placeholder.\")"
)

# If private is True, the notification protocol will only notify the action taker;
# otherwise, the change will be broadcast to all team members.
```

To support asynchronous interaction between agents, humans, and task environments, we use [Redis](https://redis.io/) for cross-process communication, and wrap each environment class with [`TaskEnvNode`](collaborative_gym/nodes/task_env.py). See [`collaborative_gym/nodes`](collaborative_gym/nodes) for other supported nodes (i.e., agents, real humans, simulated humans).

We provide a [`Runner`](collaborative_gym/runner.py) class to handle the lifecycle of human-agent collaboration sessions, including launching these node processes and ensuring proper cleanup on exit.

## Simulated and Real Conditions
Co-Gym supports two experimental conditions: Co-Gym (Simulated) and Co-Gym (Real)
- Co-Gym (Simulated) experiments with simulated humans and pre-collected task instances, allowing controlled, iterative development.
- Co-Gym (Real) experiments with real humans and user-provided task instances, allowing deploying and evaluating collaborative agents in the wild.

### Co-Gym (Simulated)
We use [TravelPlanner](https://github.com/OSU-NLP-Group/TravelPlanner) subset, arXiv CS paper subset curated in this work, [DiscoveryBench](https://github.com/allenai/discoverybench) subset for the simulated experiments on Travel Planning, Related Work, Tabular Analysis tasks respectively. See [`datasets/README.md`](datasets/README.md) for dataset details.

We use `gpt-4o` to simulate human behavior in our paper.

To reproduce our paper experiments:
1. Start Redis server: `docker run -d --name redis-stack -p 6379:6379 -p 8001:8001 redis/redis-stack:latest`
    - If you prefer to use Redis without the docker container (may need `sudo` permission):
      ```shell
      apt-get install redis
      systemctl start redis-server
      ```
2. Run the following command to test Fully Autonomous Agent:
    ```shell
    python -m scripts.fully_auto_agent_exp \
      --task {"travel_planning" or "related_work" or "tabular_analysis"} \
      --start-idx {start_idx_of_simualted_dataset} \
      --end-idx {end_idx_of_simulated_dataset} \
      --team-member-config-path {configs/teams/auto_agent_team_config_xxx.toml} \
      --result-dir-tag {result_dir_tag}
    ```
    - The result will be saved in `work_dir/{task}/{result-dir-tag}/results`.
3. Run the following command to test the human-agent team:
    ```shell
    python -m scripts.collaborative_agent_exp \
      --task {"travel_planning" or "related_work" or "tabular_analysis"} \
      --start-idx {start_idx_of_simualted_dataset} \
      --end-idx {end_idx_of_simulated_dataset} \
      --team-member-config-path {"configs/teams/basic_coagent_simulated_user_team_config_xxx.toml" or "configs/teams/coagent_with_situational_planning_simulated_user_team_config_xxx.toml"} \
      --result-dir-tag {result_dir_tag}
    ```
    - The result will be saved in `work_dir/{task}/{result-dir-tag}/results`.

### Co-Gym (Real)
**You can try out collaborative agents directly through our [live research preview](https://cogym.saltlab.stanford.edu/). Each session randomly pairs you with collaborative agent. Help us evaluate different agents by sharing your ratings and feedback - and discover some surprises along the way!**

We've open-sourced our collaborative interface in `frontend/workbench`. For setup instructions, please refer to [frontend/workbench/README.md](frontend/workbench/README.md).

## Evaluation
Co-Gym supports the evaluation of collaborative agents across both collaboration outcomes and processes.

### Evaluating Collaboration Outcome
We assess the collaboration outcome with both Delivery Rate and Task Performance (requires specifying a task-specific scoring function to evaluate the outcome).

Co-Gym defines the **Collab**orative **Score** to jointly considering outcome delivery and quality:
```math
\text{Collab Score} = 1_{\text{Delivered}}\times\text{Task Performance}
```

To see the aggregated results, run the following command:
```shell
python scripts/report_simulated_result.py --result-dir {result_dir_that_include_result_folder_for_each_instance}
```

### Auditing Collaboration Process
Co-Gym analyzes the collaboration process along the following dimensions:
- **Initiative Entropy:** This metric quantifies the distribution of initiative in the human-agent team, where a uniform distribution results in a high score (with a maximum of 1) and a skewed distribution results in a low score.
  - To compute, run: `python -m collaborative_gym.eval.initiative_analysis --result-dir {result_dir_that_include_result_folder_for_each_instance}`
- **Controlled Autonomy:** We measure this dimension by (1) counting the agent's confirmation questions that effectively elicit a human response and (2) counting instances where the human verbally intervenes to halt the agent's actions.
  - To compute, run: `python -m collaborative_gym.eval.controlled_autonomy --result-dir {result_dir_that_include_result_folder_for_each_instance}`


## Add a New Agent
The current codebase supports three agents in `demo_agent/`:
- `demo_agent/auto_agent`: Fully Autonomous Agent that uses ReAct-style prompting and only considers task action space.
- `demo_agent/basic_collaborative_agent`: Collaborative Agent that uses ReAct-style prompting and considers both task action space and collaboration acts (i.e., `SendTeammateMessage`, `WaitTeammateContinue`).
- `demo_agent/collaborative_agent_with_situational_planning`: Collaborative Agent with Situational Planning which employs a two-stage decision-making approach when processing notifications. See Appendix B of [our paper](https://arxiv.org/abs/2412.15701) for details.

The simplest way to integrate an agent into Co-Gym is by ensuring compatibility with the `AgentNode` interface. `AgentNode` manages notifications from the environment and sends agent actions back via Redis, allowing you to focus on designing the agent's policy. To work with `AgentNode`, your agent should implement the following methods:
- `start(self, name: str, team_members: List[str], task_description: str, action_space: dict, example_question: str, example_trajectory: List) -> None`
  - This method is called when the task environment is initialized.
  - `name` is the name of the agent.
  - `team_members` is a list of team members in the session.
  - `task_description` is a string describing the task.
  - `action_space` is a list of actions. Each action is specified as a dictionary with the following keys:
    - `max_length`, `min_length`
    - `pattern`: Important for the agent to know what the format of the action should be.
    - `params`: Important for the agent to know what parameters are expected in the action.
    - `machine_readable_identifier`: A unique identifier for the action.
    - `human_readable_name`: A human-readable name for the action.
    - `human_readable_description`: A more detailed description of the action.
  - `example_question` is a string representing an example question for this task (if provided by the task environment).
  - `example_trajectory` is a list of Thought/Action/Updated Observation tuples representing an example trajectory for this task (if provided by the task environment). `example_question` and `example_trajectory` can help the LM agent understand the task.
- `get_action(self, observation: Dict) -> str`
  - This method is called when the agent receives a new notification from the environment.
  - `observation` is a dictionary containing `event_log` and other task-specific observations.
  - The method should return a string representing the action to be taken following the format specified in `action_space`.
- `end(self, result_dir: str) -> None`
  - This method is called when the task is ended. The agent should clean up resources and save relevant information.


The agent can be implemented using any framework of your choice. Feel free to check out our example agents for reference.

## Add a New Task Environment
The supported task environments are under `collaborative_gym/envs`. To add a new task environment:
1. Create a new file under `collaborative_gym/envs` that inherits from [`CoEnv`](collaborative_gym/core.py). Register your environment class with `@EnvFactory.register("id_for_your_task")`.
2. Configure the environment settings:
    - Define `self.task_description` in natural language.
    - Specify action space, private and public observation spaces.
    - Optionally specify `self.example_question` and `self.example_trajectory` if you want to add task demo when running demo agents.
3. Implement `self.step()` to define the environment logic.
4. Add your environment class to [collaborative_gym/envs/__init__.py](collaborative_gym/envs/__init__.py). You can now use your environment by referencing its registered ID.

To integrate your task into the user interface, see [frontend/workbench/README.md](frontend/workbench/README.md).

## Contributions
If you have any questions or suggestions, please feel free to open an issue or pull request. We welcome contributions to add more task environments & agents and improve the codebase!
- Contributed collaborative agents can be deployed on [our research preview](https://cogym.saltlab.stanford.edu/) for evaluation once we verify its functionalities.
- Besides contributing task environments directly, you can also [request a new task](https://github.com/SALT-NLP/collaborative-gym/issues/new?template=task_environment_request.md) you would like to see officially supported in Co-Gym or vote for existing requests.

Contact person: [Yijia Shao](mailto:shaoyj@stanford.edu), [Vinay Samuel](mailto:vsamuel@andrew.cmu.edu), and [Yucheng Jiang](mailto:yuchengj@stanford.edu)

## Acknowledgement
The development of Co-Gym won't be possible without these open-source projects:
- [Gymnasium](https://github.com/Farama-Foundation/Gymnasium): for defining API standard for single-agent reinforcement learning environments.
- [storm](https://github.com/stanford-oval/storm): for providing LM modules, retriever modules, and workflows for knowledge-intensive writing tasks.
- [aact](https://github.com/ProKil/aact): for supporting asynchronous communications between nodes (i.e., humans, agents, environments in this project).

We are very grateful to the following amazing designers who have contributed to this project:
- Co-Gym (Real) UI design: Yanan Zhang
- Logo design: Long Lin

## Citation
Please cite our paper if you use this code or part of it in your work:
```
@misc{shao2025collaborativegym,
      title={Collaborative Gym: A Framework for Enabling and Evaluating Human-Agent Collaboration}, 
      author={Yijia Shao and Vinay Samuel and Yucheng Jiang and John Yang and Diyi Yang},
      year={2025},
      eprint={2412.15701},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2412.15701}, 
}
```
