Sage AI CLI

1.19.1 verified Sat May 09 auth: no python

Sage is a local-first AI coding CLI (like Claude Code) that supports free/open models. Current version 1.19.1. Released regularly on GitHub.

pip install sage-ai-cli

Common errors

error ModuleNotFoundError: No module named 'sage' ↓

cause Installed sage-ai-cli but trying to import 'sage_cli' or 'sage_ai'.

fix

Use 'from sage import SageClient' (module name is 'sage').

error ImportError: cannot import name 'SageClient' from 'sage' ↓

cause Older version of sage-ai-cli where SageClient was named 'Sage' or not exported.

fix

Upgrade sage-ai-cli: pip install --upgrade sage-ai-cli

error RuntimeError: Model not found at path /path/to/model ↓

cause Local model path specified incorrectly or model not downloaded.

fix

Use absolute path and ensure model exists. Or omit 'model_path' to use default.

Warnings

breaking Version 1.19.0 dropped support for Python 3.9; requires >=3.10. ↓

fix Upgrade Python to 3.10+.

deprecated The method `SageClient.generate_text()` is deprecated. Use `generate()` instead. ↓

fix Replace generate_text() with generate() in your code.

gotcha API key is not required for local models, but if set it overrides local inference. Many users mistakenly set a key when intending to use local models. ↓

fix Do not set SAGE_API_KEY if you want to use local models; leave it unset.

Imports

sage

wrong

from sage_ai import Sage

correct

from sage import SageClient

Old import path; correct module is 'sage'

SageClient
wrong
```
from sage_cli import SageClient
```
correct
```
from sage import SageClient
```
Common mistake: using hyphenated module name

Quickstart

Initialize SageClient with API key and generate a response.

from sage import SageClient
from os import environ

client = SageClient(api_key=environ.get('SAGE_API_KEY', ''))
response = client.generate('Explain Python decorators in one sentence.')
print(response['text'])