Getting Started

Installation

pip install operon-ai
        

For provider-backed workflows, configure the model backend you want to use through the Nucleus provider layer.

Recommended First Steps

1. Start with the Pattern-First API.
2. Run the pattern-first example and the skill-organism example.
3. Read the Examples page for the shortest path into the rest of the library.

First Practical Questions

If you are evaluating Operon, the right first questions are usually:

• should this stay single-agent?
• does it need a reviewer bottleneck?
• is it actually decomposable into specialists?
• should fixed work go to a cheap model and fuzzy work to a stronger one?

That is why the current front door is pattern-first rather than theory-first.

Minimal Entry Points

• advise_topology(...) for architecture guidance
• reviewer_gate(...) for one-worker-plus-reviewer
• specialist_swarm(...) for centralized specialist decomposition
• skill_organism(...) for multi-stage workflows with attachable components
• managed_organism(...) for the full stack in one call (adaptive assembly, watcher, substrate, development, social learning). Requires either a seeded library with templates or explicit stages=.