How to use Codex to revive a React Redux app

The author revived Highlander-react-redux, a nine-year-old application, using Turtle AI, a repository-scoped workflow built around Codex. The goal was to preserve the existing product while improving architecture, usability, and deployment rather than replacing the stack.

Work began with repository analysis and a written plan for one feature. Each implementation step passed through verification, an engineer checkpoint, testing, debugging, and a plan update. The checkpoint required the developer to explain the change, its rationale, its place in the system, and remaining risks before proceeding. Dedicated skills also covered security, performance, documentation, and backlog management.

The shipped product gained derived player analytics, season-aware teams and statistics, game-based stat entry, search and filtering, and a clearer demo experience. Production work included a multi-stage Docker build, Fly.io configuration, HTTPS enforcement, a health endpoint, PostgreSQL-backed sessions, release migrations, and demo seed data.

The deployed application serves its React build and Express API from the same origin. The author reports that the workflow helped maintain understanding of generated code, while noting that additional guardrails do not automatically improve efficiency. Who it's for: developers modernizing a legacy application who want incremental implementation with explicit comprehension and verification checkpoints.

Prompt
Suggested prompt (written by Agents In Use, not the author's)

Source: dev.to · heriberto_codes