Can't pass TavilySearchResults api key via argument in a Jupyter notebook

[smmills]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

from langchain_community.tools.tavily_search import TavilySearchResults

tavily_api_key = "tvly-xxxx"
search = TavilySearchResults(tavily_api_key=tavily_api_key)

Description

The above code works fine if run in interactive/interpret python on the command line, but does not work in a jupyter notebook instead resulting in a

ValidationError: 1 validation error for TavilySearchAPIWrapper
__root__
  Did not find tavily_api_key, please add an environment variable `TAVILY_API_KEY` which contains it, or pass `tavily_api_key` as a named parameter. (type=value_error)

Using

jupyter_client==8.6.1
jupyter_core==5.7.2
jupyter_server==2.13.0

System Info

System Information
------------------
> OS:  Darwin
> OS Version:  Darwin Kernel Version 22.6.0: Wed Jul  5 22:21:56 PDT 2023; root:xnu-8796.141.3~6/RELEASE_X86_64
> Python Version:  3.11.6 (main, Nov  2 2023, 04:52:24) [Clang 14.0.3 (clang-1403.0.22.14.1)]

Package Information
-------------------
> langchain_core: 0.1.42
> langchain: 0.1.16
> langchain_community: 0.0.32
> langsmith: 0.1.47
> langchain_anthropic: 0.1.8
> langchain_cohere: 0.1.2
> langchain_openai: 0.1.3
> langchain_text_splitters: 0.0.1
> langchainhub: 0.1.15

Packages not installed (Not Necessarily a Problem)
--------------------------------------------------
The following packages were not found:

> langgraph
> langserve

[dosubot[bot]]

It looks like the issue with passing the tavily_api_key in a Jupyter notebook stems from how environment variables are accessed differently in Jupyter compared to a standard command line Python session. The TavilySearchAPIWrapper is designed to validate and retrieve the API key either from the passed parameters or from an environment variable named TAVILY_API_KEY. The discrepancy you're experiencing might be due to the Jupyter notebook environment not correctly accessing or recognizing the environment variable.

To address this, you can try the following steps:

1. Ensure the TAVILY_API_KEY environment variable is correctly set in your Jupyter notebook environment. You might need to restart the Jupyter kernel after setting the environment variable to ensure it's recognized.

2. Verify the syntax and correctness of how you're passing tavily_api_key as a named parameter. It's crucial to ensure there are no typos or naming discrepancies.

3. Consider debugging output by temporarily modifying the validate_environment method to print out the values dictionary and the result of attempting to retrieve the environment variable. This can help pinpoint where the retrieval process is failing.

These steps should help you resolve the issue with the TavilySearchResults in your Jupyter notebook. If the problem persists, providing more details could help in identifying a more specific solution.

Sources

• libs/community/langchain_community/utilities/tavily_search.py
• libs/langchain/langchain/utils/env.py

About Dosu

This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

If you want to continue the conversation, start your reply with @dosu-bot.

[sciarrilli]

I had the same issue with jupyter notebooks running in vscode. I found this in the langchain docs for tavily.

import getpass
import os

os.environ["TAVILY_API_KEY"] = getpass.getpass()

TavilySearchResults().invoke("whats the weather in san francisco?")

[CullerWhale]

I also have this problem. I can correctly print out the api key before passing it. I will try to give it more time since they key was just created. However, that is just a stab in the dark...

[Shevchenkojjj]

Tried this, worked in my jupyter notebook.

import os

os.environ["TAVILY_API_KEY"] = "tvly_your_api_key"

from langchain_community.tools.tavily_search import TavilySearchResults

search = TavilySearchResults()

[AFRT927]

you need to define the environment variable strictly as "TAVILY_API_KEY"

[yjgao]

import os
os.environ["TAVILY_API_KEY"] = "tvly_your_api_key"

This work for me also I try to set the key in conda env directly, it also work on my side

conda env config vars set TAVILY_API_KEY="your-key"
conda activate envName

[smmills]

Hi @Shevchenkojjj, just like for all of the other langchain constructors that take *api_key as a keyword argument, environment variables can be used to pass credentials (within the notebook using the os library or directly via your shell). This ticket, however, is regarding passing the tavily_api_key keyword argument at class instantiation time, which you have not shown works in your jupyter environment with the above code snippet.

I am not looking for a work around (don't use jupyter being another), rather trying to find a fix for the issue that the keyword argument does not work for the tavily class. For example, langchain_cohere.ChatCohere(cohere_api_key=cohere_api_key) does indeed work as expected in jupyter.

[AnantChandra]

TavilySearchResults can't directly ingest an API key. Here's the documentation.

You must pass the key via the TavilySearchAPIWrapper.

This worked for me:

from langchain_community.tools.tavily_search import TavilySearchResults
from langchain.utilities.tavily_search import TavilySearchAPIWrapper

TAVILY_API_KEY="tvly-123"
tavilySearchAPIWrapper = TavilySearchAPIWrapper(tavily_api_key=TAVILY_API_KEY)
search = TavilySearchResults(api_wrapper=tavilySearchAPIWrapper)

I share the dislike of exporting environment variables in Jupyter, hence went ahead with this approach.

[duonghuuphuc]

This approach is also workable for LangChain v0.2.