Write a plan.
Walk away.
The autonomous Claude Code loop for plan execution and multi-agent code review. Each task runs in a fresh session, keeping Claude sharp from start to finish.
$ ralphex docs/plans/add-auth.md
[09:14:22] Task 1: JWT middleware → tests pass → committed
[09:21:08] Task 2: login endpoint → tests pass → committed
[09:28:45] Review 1: 5 agents → 2 issues fixed → committed
[09:35:12] Codex: GPT-5 review → 1 issue reported
[09:41:03] Review 2: 1 fix → committed → no issues
[09:41:17] COMPLETED → docs/plans/completed/
What is ralphex?
An autonomous coding loop for Claude Code that takes the ralph loop concept further. It reads an implementation plan, executes tasks one by one in fresh Claude sessions, then runs multi-agent code review with five specialized AI agents in parallel. Unlike a simple ralph loop that re-runs the same prompt, ralphex manages structured plans, commits after each task, and orchestrates multi-phase reviews including optional GPT-5 cross-validation.