ChatGPT Danger Mode for Jupyter
Project description
🚨 Dangermode
Overview
🚨 Dangermode is a ChatGPT Plugin written with Python and FastAPI that allows ChatGPT to execute code snippets in an IPython session, whether it's the console, the notebook, or a JupyterLab session.
⚠️ Dangermode is a proof-of-concept and should not be used in production. It is not secure and should not be publicly exposed to the web. ⚠️
🚨🚨🚨 YOU ARE GIVING AI THE ABILITY TO EXECUTE CODE AS YOU. 🚨🚨🚨
Running this plugin on your machine is dangerous. Running this plugin on a public server is extremely dangerous. You have been warned.
Demo
Features
- Execute code snippets in an IPython session through ChatGPT.
- Inspect variables and view results in real-time.
- Serve images and display data to ChatGPT.
Get started, the (slightly) less dangerous way
Since you don't want to unleash ChatGPT directly onto your literal machine (including your files) etc. Build the docker image and run it locally like this:
docker build . -t dangermode
docker run -p 8000:8000 -i -t --rm dangermode
(SCARY DANGER MODE) Installation
If you're feeling brave , you can install dangermode
directly via pip
, conda
, or clone the repository and install it locally. If you really aren't worried about security, go for it. You have been warned.
Run Danger Mode
import dangermode
# You must set the host to bind to all addresses when using Docker.
# Since this is dangerous, I leave it as an exercise to the reader.
dangermode.activate_dangermode()
Enabling on ChatGPT
In order to use this plugin, you have to have ChatGPT Plugin access.
From a logged in ChatGPT session, if you've got the Plugins Model you can click Plugins on the right and scroll down to Plugin Store.
Next, click "Develop your own plugin".
Enter localhost:8000
as the domain.
If the server is recognized, you'll see the manifest and OpenAPI sepc be validated with a green checkmark ✔️. Click "Install localhost plugin" and start using it!
API Endpoints
GET /openapi.json
: Retrieve the OpenAPI JSON configuration.GET /.well-known/ai-plugin.json
: Retrieve the AI plugin JSON configuration.GET /images/{image_name}
: Retrieve an image by its name.GET /api/variable/{variable_name}
: Retrieve the value of a variable by its name.POST /api/run_cell
: Execute a code cell and return the result.
Contributing
Please do. I can't let Large Language Models write all of it.
Please fork the repository, make your changes, and submit a pull request.
License
Dangermode is released under the BSD 3-Clause License. See LICENSE for more information.
Contact
If you have any questions or feedback, please feel free to reach out to the author, Kyle Kelley, on Twitter at @KyleRayKelley or just open an issue on the repository.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for dangermode-0.3.0-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | f278a143e6e2fcbf64935acf852ac55ccbf54c98803d15d660fc00c97145b0e7 |
|
MD5 | db2e714043a8ff8b167366e5143ee965 |
|
BLAKE2b-256 | f962d4775cbc86ee8ba5eee5d6db1bbe0b6a85d0693bc19288e5982c00971baa |