A modern command line assistant.
curl https://raw.githubusercontent.com/OpenInterpreter/open-interpreter/refs/heads/development/installers/new-installer.sh | sh
Start a conversation:
$ interpreter
> help me set up a new fastapi project
Creating project structure...
Added requirements.txt, main.py, and Dockerfile
> add a /users endpoint
Adding routes/users.py...
Instant chats with i [prompt]
:
$ i want a venv here
$ i want to undo the last commit
$ i need deno
Fix errors with wtf
:
$ python test.py
ImportError: No module named 'requests'
$ wtf
Installing requests...
Done. Try running your script again.
# Choose your model
interpreter --model gpt-4o
interpreter --model claude-3-sonnet
# Save configurations
interpreter --model gpt-4o --save-profile 4o
interpreter --profile 4o
# Enable tools (default: interpreter,editor)
interpreter --tools interpreter,editor,gui
pip install open-interpreter
from interpreter import Interpreter
# Start interpreter
interpreter = Interpreter()
# Multiple tasks in same context
messages = interpreter.respond("write a test for this function")
messages = interpreter.respond("now add error handling")
# Reset context
interpreter.messages = []
# Set custom context
interpreter.messages = [{"role": "user", "content": "write a test for this function"}]
# Stream output
for chunk in interpreter.respond(stream=True):
print(chunk, end="")
# Start an interactive chat
interpreter.chat()
# View conversation history
print(interpreter.messages)
(Coming soon)