Running Six Codex Accounts, Claude Code, and Antigravity on One Mac

I run six Codex accounts using gpt 6.1 sol, alongside Claude Code, Antigravity, and Cursor. Each has a separate macOS user, with the users connected through an ai group. My live trading code has hard read only protections against the coding agents. The more useful story is about operating systems and coordination, not only about models. Account identity should stay separate from task identity. The shared group should have a narrow purpose. Work should be divided by interfaces, with an explicit integration owner for each task. Separate checkouts help prevent agents from changing the same uncommitted files, but they should not be confused with security. Handoffs should carry inspectable evidence, including changed files, checks performed, and unresolved assumptions. Availability tracking is not a capacity promise. More concurrent sessions can produce more candidate work than can be comfortably inspected, so the review queue should limit active implementation.

Authors

Publication Details

Journal
Figshare
Published
2026-11-12
DOI
https://doi.org/10.6084/m9.figshare.34993603
Primary Topic
Software Engineering Techniques and Practices
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Running Six Codex Accounts, Claude Code, and Antigravity on One Mac

Jeremy Koski
Figshare
Software Engineering Techniques and Practices
preprint

Running Six Codex Accounts, Claude Code, and Antigravity on One Mac

Jeremy Koski
preprint en

Abstract

I run six Codex accounts using gpt 6.1 sol, alongside Claude Code, Antigravity, and Cursor. Each has a separate macOS user, with the users connected through an ai group. My live trading code has hard read only protections against the coding agents. The more useful story is about operating systems and coordination, not only about models. Account identity should stay separate from task identity. The shared group should have a narrow purpose. Work should be divided by interfaces, with an explicit integration owner for each task. Separate checkouts help prevent agents from changing the same uncommitted files, but they should not be confused with security. Handoffs should carry inspectable evidence, including changed files, checks performed, and unresolved assumptions. Availability tracking is not a capacity promise. More concurrent sessions can produce more candidate work than can be comfortably inspected, so the review queue should limit active implementation.

Figshare
Software Engineering Techniques and Practices
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon