Integrate APIs with OpenAPI Specs

The OpenAPI Tool loads an OpenAPI spec and lets an agent retrieve its endpoints and endpoint details. It exposes load_openapi_spec, which returns the parsed OpenAPI spec the class was initialized with. Pairing it with the companion RequestsToolSpec lets the agent actually hit the endpoints it discovers with a REST request, rather than just reading about them.

Use it when an agent needs to explore or reason about an API's shape -- endpoints, parameters, base URL -- from its OpenAPI spec, optionally following up with real requests via RequestsToolSpec. On its own it only reads the spec; without adding RequestsToolSpec to the agent's tools, it cannot actually call the API.

Install with:

pip install llama-index-tools-openapi

Initialize OpenAPIToolSpec either with a URL to the spec, or with an already-parsed spec dict:

from llama_index.tools.openapi import OpenAPIToolSpec
from llama_index.core.agent.workflow import FunctionAgent
from llama_index.llms.openai import OpenAI

f = requests.get(
    "https://raw.githubusercontent.com/APIs-guru/openapi-directory/main/APIs/openai.com/1.2.0/openapi.yaml"
).text
open_api_spec = yaml.safe_load(f)
### OR
open_spec = OpenAPIToolSpec(
    url="https://raw.githubusercontent.com/APIs-guru/openapi-directory/main/APIs/openai.com/1.2.0/openapi.yaml"
)


tool_spec = OpenAPIToolSpec(open_api_spec)

agent = FunctionAgent(
    tools=tool_spec.to_tool_list(),
    llm=OpenAI(model="gpt-4.1"),
)

print(await agent.run("What is the base url for the API"))
print(await agent.run("What parameters does the x endpoint need?"))

Developers building agents that need to understand and navigate an API's OpenAPI spec -- and optionally call its endpoints -- without manually reading API documentation.