Create a Lower-Cost Luna Worker for Codex Subagents

the useful split

Keep Sol on decisions. Send bounded work to Luna Max.

A stronger orchestrator handles the parts where context and judgment matter. A lower-cost worker handles small jobs with explicit scope and proof.

Sol01

orchestrates

Luna Max02

implements

Checks03

prove

why test the split

Lower cost is useful when the task has edges

Luna versus Sol benchmark chart showing intelligence index score against cost per task — Supplied Luna versus Sol Intelligence Index chart: Luna Max is shown near Sol Medium on the measured score at a lower task cost. Treat the comparison as benchmark-specific, not universal equivalence.

copy this into Codex

Create the worker safely

This original instruction makes Codex inspect the installed setup, preserve it, show the diff, validate compatibility, and only then use the worker.

codex-luna-worker.txt

create a custom agent named luna_worker at ~/.codex/agents/luna-worker.toml.
use these settings:
model = "gpt-5.6-luna"
model_reasoning_effort = "max"
give it a description and instructions for bounded delegated work. preserve the rest of my config. validate it against my installed Codex version, show me the diff, then use luna_worker for subagent tasks.

keep with Sol

Judgment

Architecture and technical trade-offs
Ambiguous or changing requirements
Safety, security, and high-impact changes

send to Luna Max

Throughput

One named function, component, migration, or test
A clear file or folder boundary
A check that proves the task is done

The caveat

Use the stronger model for architecture, ambiguity, safety, and high-impact changes. Use Luna Max for concrete, bounded, verifiable work. The check at the handoff is what makes the saving useful.

Orchestratedelegateverify