Custom python package installation

[sbhonagiri-evoke]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
• Please do not modify this template :) and fill in all the required fields.

1. Is this request related to a challenge you're experiencing?

i have to run a separate python endpoint since dify doesn't has the ability

2. Describe the feature you'd like to see

i just need an option to enter the pip package name and it should be installed in the backend to use in code node

3. How will this feature improve your workflow or experience?

this will improve the overall user experience and reduce lot of manual work

4. Additional context or comments

No response

5. Can you help us with this feature?

• I am interested in contributing to this feature.

[huy302]

I'm interested in this feature as well, Dify would be so much more useful if we are allowed to install additional python packages to run in the code block

[EchterTimo]

@sbhonagiri-evoke
@huy302

Little late but I found what you need:

Warning

This only works if you are on a selfhosted instance. You can not add custom packages if you are using cloud.dify.ai

In my case I am on a docker based deployment: https://docs.dify.ai/en/self-host/quick-start/docker-compose

Steps

1. create file /docker/dependencies/python-requirements.txt
2. Enter Python libs in that file like you do with requirements.txt. Versioning your packages is advised if you don't want new releases breaking your code.

dify-client==0.1.10

Warning

Libs like numpy need extra SYSCALLS permissions that you need to specify. This is mentioned by the lead maintainer in this discussion: #8200 (comment)
However if you are just using an api wrapper like dify-client you should be fine without extra permissions.
Feel free to try other libs and share your experience.

Warning

The default Sanbox Python version is 3.10.17.
Make sure the libs you want to use are compatible with that version.
You can check the version by using a code node with the following code.

import sys
def main():
    return {
        "version": sys.version
    }

Note

Here you can find out what dependencies the dify sandbox already installs for you:
https://github.com/langgenius/dify-sandbox/blob/3f563d94d9ba2201c495f5653e2830745331c478/docker/templates/production.dockerfile#L4

Currently these are: httpx, requests, jinja2, PySocks, httpx[socks]

3. Finally run this command to restart the sandbox

sudo docker restart docker-sandbox-1

4. Verify the container is running by using this command:

docker ps -a | grep -e sandbox -e "CONTAINER ID"

5. Check the logs of the sandbox container to see if the dependencies were installed correctly. You can use this command to see the last 300 lines of logs:

docker compose logs --tail=300 sandbox

You want to have the following line in the logs:

sandbox-1  | 2026/02/19 20:56:51 server.go:60: [INFO]python dependencies sandbox initialized

6. You should now be able to use the libs in the sandbox

Where do I have this information from?

• Dify Sandbox v0.2.6: Install any Python package you want #8200
• https://forum.dify.ai/t/how-can-i-install-python-packages-in-the-code-node/55
• https://github.com/langgenius/dify-sandbox/blob/main/FAQ.md
• testing it out myself