🏭 ToolFactory: Convert Any REST API Documentation into AI-Usable Tools 🏭

ToolFactory is your one-stop solution for turning messy, unstructured REST API documentation into polished, AI-compatible tools. Built for developers, by developers, it’s designed to take the hassle out of integrating APIs into LLM-based tool agents. Say goodbye to manually parsing through inconsistent docs and hello to a streamlined pipeline for building natural language-enabled tools.

🏭 What is ToolFactory? 🏭

APIs are the backbone of modern applications, but their documentation? Not so much. ToolFactory automates the transformation of REST API docs—no matter how unstructured—into tools ready to plug into AI workflows. It’s built to make your life easier and your tools smarter.

• 🛠 Automated Tool Generation Convert unstructured API docs into AI-ready tools in minutes. No sweat, no hassle.

• 🩺 Diagnosing Diagnose and squash errors in your generated tools with our evaluation method.

• 📚 Knowledge Base Magic Leverage a curated library of verified tools to fill gaps in poorly documented APIs.

🛠 How It Works

1. Feed the Machine: Supply your unstructured REST API docs.
2. Let ToolFactory Do Its Thing: Watch as it parses, diagnoses, and transforms APIs into tools.
3. Integrate and Dominate: Plug the tools into your LLM-based agents and level up your AI workflows.

Contributions welcome. Issues encouraged. Stars appreciated. ⭐ Let's make API integration awesome together.

Get Started

Setup environment

Open .env and enter your OPENAI_API_KEY.

pip install -r requirements.txt

Prepare Your Documents

Create a new folder under extractor\apidocs\ and rename it as your API name, for example, extractor\apidocs\exampleapi. Use the "save page as" button in your browser to download the web html.

Generate Tools

python extractor\api_code_generator.py extractor\apidocs

Occasionally, the generated tools will have no base url(not specified in the raw documentation). In this case, you can manually add a .config file in the API folder with

{
    "base_url": "https://your_base_url_here"
}

Make sure you use http:// or https:// in the base url, otherwise it will fail in the validation. After that, overwrite the tools

python extractor\api_code_generator.py extractor\apidocs -o

Validation

Check main\tool_testing.ipynb for more details.

Export

Export your tools into openapi specification.

python post_process\generate_openapi.py extractor\apidocs

The output yaml specification will be in the api folder.

For research

API Extraction Benchmark available on huggingface billyfin\APIdoc2json Our prompt tuned APILLAMA can be accessed at billyfin\APILlama Our training and evaluation code under APILlama folder

We recommend to use a stand-alone environment for training and testing APILlama. An independent requirements.txt is provided.