Capability · Framework — orchestration

ell

ell (ell.so) rethinks prompt engineering as software engineering. Prompts are Python functions decorated with @ell.simple or @ell.complex; every call is versioned, diffed, and logged to a local SQLite store. A Studio UI lets you browse the history, compare versions, and rerun past prompts.

Framework facts

Category: orchestration
Language: Python
License: MIT
Repository: https://github.com/MadcowD/ell

Install

pip install -U ell-ai

Quickstart

import ell

ell.init(store='./ell_store', autocommit=True)

@ell.simple(model='claude-opus-4-7')
def elevator_pitch(topic: str) -> str:
    '''You are a concise marketing copywriter.'''
    return f'Write a 2-sentence elevator pitch about {topic}.'

print(elevator_pitch('VSET — a Delhi engineering school'))

Alternatives

DSPy — programmatic prompt optimisation
Mirascope — Pythonic prompt classes
Instructor — structured outputs
Promptflow — Microsoft flow authoring

Frequently asked questions

Why would I use ell over plain API calls?

Because ell automatically versions prompts and logs every invocation, giving you Git-like history over prompt iterations without any extra wiring.

Is ell production-ready?

It's maturing fast and used in real products. For enterprise-grade tracing most teams pair ell with Langfuse or LangSmith on top of the built-in store.

Sources

ell — GitHub — accessed 2026-04-20
ell — docs — accessed 2026-04-20