
Voting open · ends in—
Secure multiple Codex accounts, local and remote
CodexAccountBridge (cab) is an open-source Codex account manager for macOS and Linux. It provides a command-line tool and a native SwiftUI app for managing multiple official Codex or ChatGPT logins across local machines, SSH servers, remote projects, and supported proxy services.
Each login is stored in its own CODEX_HOME, with authentication and model traffic delegated through the official Codex executable. CAB does not implement OAuth, proxy model requests, inspect token contents, or silently switch accounts after quota limits or errors.
Conservative security model: each account logs in independently and never reads or copies auth.json.
Key features include: account management, usage limit viewing, macOS notifications, SSH server management, session isolation, and service binding.
Product Keywords
Loading comments...
Maker
CodexAccountBridge (cab) is an open-source Codex account manager for macOS and Linux. It combines a command-line tool with a native SwiftUI app to help you juggle multiple official Codex or ChatGPT logins across local machines, SSH servers, remote projects, and supported agent services. Each login lives in its own isolated CODEX_HOME, while authentication and model traffic flow through the official Codex executable — no shortcuts, no credential merging.
Create, list, and switch between independent Codex accounts with simple commands like cab account add, cab use, and cab run. Removing a registration never deletes the underlying account directory, login, or session history.
Check official Codex usage limits and reset windows for each account at a glance. Normal usage reads are display-only, and the optional usage-period wake is an explicit, opt-in minimal request — never an automatic rotation trigger.
Manage accounts on multiple SSH servers directly from the native macOS app. This extends your multi-account workflow beyond local machines to remote infrastructure without leaving the desktop interface.
Preserve or isolate project and session history during deliberate account changes, with opt-in cross-account session sharing that excludes credentials and SQLite state. You can also bind supported Hermes and OpenClaw systemd user services to a chosen Codex account and configure a visible launch rotation order.
Conservative security model: each account logs in independently and never reads or copies auth.json.
CAB deliberately avoids implementing OAuth, proxying model requests, inspecting token contents, or reacting to quota limits and errors. The official codex executable is invoked directly — no shell, no argument rewriting — and Codex approval, sandbox, and project-trust defaults stay untouched. Configuration updates are locked, atomic, permission-restricted, and recoverable from backups, making this one of the most boundary-respecting account managers available.
You run multiple Codex or ChatGPT logins and want strict credential separation with a native macOS experience, or you manage Codex across SSH servers and remote projects and need explicit account control without automation that silently switches identities. It's also a strong fit if you value a security model that treats auth.json as untouchable and keeps every login flow official.
Other tools you might consider
Unsloth Desktop is an open-source app to run and train AI models locally. Run LLMs, image/video diffusion, and audio. Connect agents like Claude Code or Codex to your local GPU with one command, and fine-tune models with no-code workflows.
Capsomnia turns Caps Lock into a physical switch that keeps your MacBook awake. Keep Claude Code, Codex, other AI agents, SSH sessions, builds, downloads, and more running after you close the lid. If Caps Lock is already assigned, choose another shortcut. The Caps Lock light still shows whether Capsomnia is active. The app is free and open source, signed and notarized. No accounts or telemetry. We’ve already seen strong interest in Japan: 2.3M+ views and 6,000+ downloads.
You can now give Hermes, Claude Code, and Codex infinite memory. Agentmemory is trending on GitHub with 5,000+ Stars. CLAUDE md dumps 22,000+ tokens into context at 240 observations agentmemory: 1,900 tokens. same observations. 92% less. At 1,000 observations, 80% of your built-in memories become invisible. agentmemory keeps 100% searchable. benchmarked on 240 real coding sessions → Up to 95% fewer tokens per session → 200x more tool calls before hitting context limits → 100% open source
Pi is a minimal terminal coding harness. Adapt Pi to your workflows, not the other way around. Customize Pi with extensions, skills, prompt templates, and themes. Bundle them as Pi packages and share via npm or git. Pi ships with powerful defaults but skips features like sub-agents and plan mode. Ask Pi to build what you want, or install a package that does it your way.
Loading comments...