LangChain Exa Integration

1.1.0 verified Fri May 01 auth: no python

An integration package connecting Exa (formerly Metaphor) and LangChain. Current version 1.1.0, requires Python >=3.10 and <4.0.0. Maintained under the LangChain monorepo, with releases tied to LangChain core updates.

pip install langchain-exa

Common errors

error ModuleNotFoundError: No module named 'langchain_exa' ↓

cause Package not installed or imported with wrong name (e.g., 'langchain-exa' instead of 'langchain_exa').

fix

Run 'pip install langchain-exa' and use 'from langchain_exa import ExaSearchToolkit'.

error ImportError: cannot import name 'ExaSearchToolkit' from 'langchain_exa.tools' ↓

cause Incorrect import path; ExaSearchToolkit is at the root package.

fix

Use 'from langchain_exa import ExaSearchToolkit'.

error exa_py.exceptions.AuthenticationError: API key is invalid or missing ↓

cause No valid EXA_API_KEY provided.

fix

Set the EXA_API_KEY environment variable or pass api_key='your-key' to the constructor.

Warnings

breaking In v1.0.0, the package was renamed from 'langchain-metaphor' to 'langchain-exa'. All imports have changed. ↓

fix Replace 'from langchain_metaphor import ...' with 'from langchain_exa import ...'.

deprecated The 'MetaphorSearchToolkit' alias was removed in v1.1.0. ↓

fix Use 'ExaSearchToolkit' instead.

gotcha If you pass the API key directly, it may be logged or exposed in traces. Always use environment variables or a secrets manager in production. ↓

fix Set EXA_API_KEY via os.environ and omit the api_key parameter.

Imports

ExaSearchToolkit

wrong

from langchain_exa.tools import ExaSearchToolkit

correct

from langchain_exa import ExaSearchToolkit

ExaSearchToolkit is exposed at the package root, not in a submodule.

ExaRetriever
```
from langchain_exa import ExaRetriever
```

Quickstart

Initialize the Exa search toolkit with an API key and perform a search.

import os
from langchain_exa import ExaSearchToolkit

api_key = os.environ.get('EXA_API_KEY', '')
toolkit = ExaSearchToolkit(api_key=api_key, max_results=5)
result = toolkit.search('latest AI news')
print(result)