GitHub radar
Codex Multi-Agent Setup: Astra Orchestrates, Luna Executes
A ready-made Codex configuration that splits work between GPT-6 Astra as planner-reviewer and GPT-5.6 Luna as executor—multi-agent in one install script.
A configuration package for Codex that sets up a multi-agent topology: GPT-6 Astra acts as root orchestrator and final reviewer, while GPT-5.6 Luna handles execution subagent work. Includes five agent roles—explorer, worker, researcher, tester, and reviewer—plus a token-usage analysis tool and install scripts for macOS, Linux, and Windows. Adapts automatically to Pro or Plus plans during setup, keeping orchestrated sessions within Plus rate-limit windows when needed. The Astra orchestrator skill is invoked with $astra-orchestrator.
Why a vibe-coder should care
Every task gets a final review by a stronger model without extra manual steps. Execution is routed to the faster subagent model, so you get both quality control and token efficiency in the same session.
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Install the configuration from https://github.com/donvito/codex-astra-luna-orchestrator — follow the README, ask me my subscription plan (Pro or Plus), set up the multi-agent $astra-orchestrator mode and show me how to run it
▌ More finds