Tutorial

This tutorial covers every major component of russo in detail. Work through it sequentially, or jump to the section you need.

Overview

russo's architecture is a simple pipeline with pluggable components:

graph LR
    A["Text Prompt"] --> B["Synthesizer"]
    B -->|Audio| C["Agent"]
    C -->|AgentResponse| D["Evaluator"]
    D -->|EvalResult| E["Pass / Fail"]

Each component is defined by a protocol (structural typing). You don't need to inherit from anything — just implement the right method signature.

Sections

Section	What You'll Learn
Pipeline	How russo.run() orchestrates the full flow
Adapters	Connecting to Gemini, OpenAI, HTTP, WebSocket, or custom agents
Synthesizers	Converting text prompts to audio
Evaluators	Comparing expected vs actual tool calls
Caching	Skipping TTS calls on repeated test runs
pytest Plugin	Markers, fixtures, CLI options, and reporting