---
title: "Introducing Roast: Structured AI workflows made easy"
description: "Shopify’s Augmented Engineering Developer Experience team built and open-sourced Roast to make AI-assisted developer workflows more reliable than free-form agents operating across large codebases. Roast is a convention-oriented orchestration framework that combines non-deterministic AI behavior with ordinary code execution through YAML workflows, Markdown prompts, shell commands, Ruby steps, and parallel or conditional control flow. Its shared context, built-in tools, Claude Code integration through CodingAgent, and automatic session replay support workflows that are version-controlled, testable, resumable, and adaptive within defined guardrails. Shopify reports using it to analyze test quality at scale, add Sorbet annotations through the Boba workflow, aggregate competitive intelligence, and research code history, while the project is distributed as a Ruby gem and supports AI providers through Raix."
---

# Introducing Roast: Structured AI workflows made easy

[Shopify](https://yomu.fyi/company/shopify) · 2023-10-18 · Jun 18, 2025

**Type:** Announcement

## Summary

Shopify’s Augmented Engineering Developer Experience team built and open-sourced Roast to make AI-assisted developer workflows more reliable than free-form agents operating across large codebases. Roast is a convention-oriented orchestration framework that combines non-deterministic AI behavior with ordinary code execution through YAML workflows, Markdown prompts, shell commands, Ruby steps, and parallel or conditional control flow. Its shared context, built-in tools, Claude Code integration through CodingAgent, and automatic session replay support workflows that are version-controlled, testable, resumable, and adaptive within defined guardrails. Shopify reports using it to analyze test quality at scale, add Sorbet annotations through the Boba workflow, aggregate competitive intelligence, and research code history, while the project is distributed as a Ruby gem and supports AI providers through Raix.

## Context

The team found that AI agents were unreliable when allowed to operate freely across large codebases. Complicated prompts worked better when divided into discrete steps, and non-determinism created challenges for reliable developer-productivity workflows such as grading and optimizing unit tests at scale.

## Approach / What changed

Roast provides declarative, convention-oriented workflows built from YAML configuration and Markdown prompts. It interleaves AI and non-AI steps, including shell commands, custom Ruby classes, parallel execution, iteration, conditionals, case statements, shared conversation context, built-in file and command tools, and the CodingAgent integration with Claude Code. Executions are saved for session replay, and Raix provides provider abstraction, retries, caching, structured outputs, and configuration controls.

## Takeaways

- Roast workflows can combine directory-based prompts, command execution, inline prompts, custom Ruby steps, and nested arrays for parallel execution.
- CodingAgent gives an AI agent autonomy to iterate, test, and improve within workflow-defined objectives and guardrails; Shopify uses this in Boba after deterministic cleanup and Sorbet autocorrection.
- Automatic session replay saves every workflow execution and allows developers to resume from a chosen step instead of rerunning earlier AI operations.

**Tags:** [Developer Experience](https://yomu.fyi/topic/developer-experience), [LLMs](https://yomu.fyi/topic/llm), [Open Source](https://yomu.fyi/topic/open-source), [Testing](https://yomu.fyi/topic/testing)

- Source: [Shopify](https://shopify.engineering/introducing-roast)
- Source URL: https://shopify.engineering/introducing-roast
- Ingested by Yomu: 2026-08-30T13:19:42.789Z

[Read original post](https://shopify.engineering/introducing-roast)