Metadata-Version: 2.4
Name: shunt
Version: 0.0.1
Summary: Routes LLM requests between cheap open-source and premium models based on task difficulty
Project-URL: Homepage, https://github.com/PierrunoYT/shunt.sh
Author: PierrunoYT
License: MIT
License-File: LICENSE
Keywords: cost-optimization,llm,openai,openrouter,router
Classifier: Development Status :: 1 - Planning
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: Software Development :: Libraries
Requires-Python: >=3.9
Description-Content-Type: text/markdown

# shunt

**Early development — APIs will change.**

Shunt is a small, self-hosted router with an OpenAI-compatible API. It classifies
every request — coding, reasoning, light chat — and forwards it to the cheapest
open-source model that can handle it.

```python
from shunt import Router

router = Router()
track = router.classify("Refactor this function to be async")
# -> "coding"
```

Planned for upcoming releases:

- `shunt serve` — local OpenAI-compatible proxy server
- Ordered fallback chains per track ("light", "general", "coding", "coding_heavy", "reasoning")
- YAML configuration and a local SQLite cost ledger

## Status

`v0.0.1` is a name-holding stub for the project under active development.
Follow progress at the project repository.

## License

MIT
