Speakeasy is the fastest way to ship developer experience for your APIs.
speakeasyapi.dev

OpenAPI agent examples

This repository demonstrates how to build OpenAPI-powered AI agents using two popular frameworks, LangChain and Haystack. These agents can process user queries, interact with APIs described by OpenAPI documents, and generate intelligent responses.

For an in-depth comparison of LangChain and Haystack, see this article.

Repository structure

The repository is organized into the following directories:

1. f1-fastapi-server

A mock Formula One (F1) API server built using FastAPI. This API serves as the backend for the agents.

• Files:
  • main.py: FastAPI application defining endpoints for querying F1 race winners.
  • openapi.yaml: OpenAPI specification describing the API's endpoints, request parameters, and response formats.
  • README.md: Instructions for setting up and running the F1 API server.
  • requirements.txt: Dependencies required to run the server.

2. langchain-agent

An AI agent built with LangChain that interacts with the F1 API to fetch race data based on user queries.

• Files:
  • langchain_agent.py: Script defining the LangChain agent.
  • openapi.json: OpenAPI document (converted to JSON format) for the F1 API.
  • README.md: Instructions for setting up and running the LangChain agent.
  • requirements.txt: Dependencies required for the LangChain agent.

3. haystack-agent

An AI agent built with Haystack that performs similar tasks as the LangChain agent.

• Files:
  • haystack_agent.py: Script defining the Haystack agent.
  • openapi.yaml: OpenAPI document for the F1 API.
  • README.md: Instructions for setting up and running the Haystack agent.
  • requirements.txt: Dependencies required for the Haystack agent.

Setup instructions

Prerequisites

• Python 3.9+
• A valid API key from OpenAI or Anthropic.

Steps

1. Clone the repository:

   git clone https://github.com/ritza-co/openapi-agent-examples.git
   cd openapi-agent-examples

2. Set up the F1 API server:

   • Navigate to the f1-fastapi-server directory:

     cd f1-fastapi-server

   • Install dependencies:

     pip install -r requirements.txt

   • Run the server:

     uvicorn main:app --reload

   The server will be available at http://127.0.0.1:8000.

   To get the OpenAPI document for the API server, navigate to http://127.0.0.1:8000/openapi.json.

3. Run the LangChain agent:

   • Navigate to the langchain-agent directory:

     cd langchain-agent

   • Install dependencies:

     pip install -r requirements.txt

   • Set your Anthropic API key as an environment variable:

     export ANTHROPIC_API_KEY="your_api_key"

   • Run the agent with a query:

     python langchain_agent.py "Who won the Monaco Grand Prix in 2024?"

4. Run the Haystack agent:

   • Navigate to the haystack-agent directory:

     cd haystack-agent

   • Install dependencies:

     pip install -r requirements.txt

   • Set your OpenAI API key as an environment variable:

     export OPENAI_API_KEY="your_api_key"

   • Run the agent with a query:

     python haystack_agent.py "Who won the Monaco Grand Prix in 2024?"

Comparison of LangChain and Haystack

Both LangChain and Haystack are capable frameworks for building OpenAPI-powered AI agents. See this article for an in-depth comparison.