23 • Mobile Engineer 📱 iOS · macOS · Android Building weird ideas into useful apps. @useHubbit · @use_motivv · @trykept Open to Mobile Dev roles :)
Manchester, England
Joined April 2021
- Tweets2.8K
- Following753
- Followers886
- Likes24.9K
Akintade Oluwaseun retweeted
How it works:
Each account gets its own profile, signed in once through the agent's own login.
Run csm claude like normal. Hit a limit? Run csm next to continue on your next account with the same conversation.
Or enable csm auto automatic and CSM handles confirmed usage limits for you,
saving a checkpoint and resuming your work on the next ready account.
No logout. No copy-paste. No guessing.
Akintade Oluwaseun retweeted
Happy to launch CSM 🚀
If you use Claude Code, Codex, Gemini CLI or Copilot CLI with more than one account, CSM lets you switch between them without logging out.
Your session, conversation and context come with you. Pick up exactly where you stopped.
usecsm.xyz
Domain from @brimblehq btw✨
Akintade Oluwaseun retweeted
Tired of logging out and back in every time you need to switch Claude Code or Codex accounts?
I built CSM to make that easier.
Switch between accounts without the repeated login process, while keeping each account’s configuration and sessions separate.
GitHub: github.com/onoja123/csm
Akintade Oluwaseun retweeted
CSM Desktop is now open source. 🚀
Run Claude Code, Codex, Gemini, Copilot and Cursor in one workspace.
Track sessions in Mission Control, hand off tasks with context, and use the macOS menu bar to monitor agents, approvals and rate limits without opening the app.
Repo: github.com/onoja123/csm-desk…
Akintade Oluwaseun retweeted
The AI stylist and the mannequin are now real. Real wardrobe, real AI. No mock data.
What you see here: → Dress the mannequin and keep using the app while it's drawn in the background → A notification tells you when your look is ready → The AI stylist suggests more than one outfit, and learns what you don't like
Waits are sped up. Still early and still imperfect, but it's working end to end.
#buildinpublic
Introducing Wardrobe an AI app that saves your clothes from uploaded images and styles them for the weather, city and occasion.
Mock data for now. Flow: onboarding → name → mannequin → first piece → Home.
#buildinpublic
Didn’t win anything at Shipaton.
Still came out with an app on the App Store and six weeks of shipping in public.
Kept's still live and I'm still building it that was always the actual goal.
Big thank to @RevenueCat for the opportunity to build 🚀
Back to work 🧑🍳
Akintade Oluwaseun retweeted
Use CSM today to switch between Claude Code, Codex, Gemini, Cursor, and Copilot, manage multiple accounts, and hand off work between agents without losing context.
One workspace. Your agents. Your choice.
github.com/onoja123/csm
Akintade Oluwaseun retweeted
I had a backlog of Sentry bugs and thought, why not get CSM to fix them?
So I built AutoFix into CSM.
Connect Sentry once, and every new error kicks off a run. An agent investigates, CSM reproduces the bug in an isolated Docker sandbox, fixes it in a separate worktree, verifies the fix, and gets a second agent to review the changes.
The PR still waits for my approval. Nothing merges or deploys on its own.
github.com/onoja123/csm
Built CSM 🫡
A CLI for running multiple Claude Code, Codex, Gemini CLI & GitHub Copilot CLI accounts on one machine.
Switch accounts without logging out, keep your session, and automatically fail over to another account when one hits its usage limit.
Your tokens/cookies/Keychain secrets stay with the agents. CSM doesn’t store or copy them.
Open source → github.com/onoja123/csm
Akintade Oluwaseun retweeted
I didn’t want to use the csm desktop, so i went ahead and added Sentry to the CSM CLI so I can fix bugs without leaving my terminal.
csm issues lists unresolved issues. csm fix ISSUE-ID kicks off a run that reproduces the bug in an isolated Docker sandbox, works on a fix in a separate git worktree, runs checks, and gets a second agent to review it.
The PR waits for my approval. No automatic merges or deploys.
github.com/onoja123/csm
Built CSM 🫡
A CLI for running multiple Claude Code, Codex, Gemini CLI & GitHub Copilot CLI accounts on one machine.
Switch accounts without logging out, keep your session, and automatically fail over to another account when one hits its usage limit.
Your tokens/cookies/Keychain secrets stay with the agents. CSM doesn’t store or copy them.
Open source → github.com/onoja123/csm
Akintade Oluwaseun retweeted
CSM lets you run multiple accounts of the same coding agent and automatically switch accounts when you hit usage limits, without losing your conversation. 🥸
github.com/onoja123/csm
Built CSM 🫡
A CLI for running multiple Claude Code, Codex, Gemini CLI & GitHub Copilot CLI accounts on one machine.
Switch accounts without logging out, keep your session, and automatically fail over to another account when one hits its usage limit.
Your tokens/cookies/Keychain secrets stay with the agents. CSM doesn’t store or copy them.
Open source → github.com/onoja123/csm
Akintade Oluwaseun retweeted
AI agents can say “all tests passed.” I wanted CSM to show the evidence.
So I built Build Receipts into CSM, because I think we should be able to trust the work, not just the agent’s word.
See which commands actually ran, what passed or failed, which checks were skipped, and exactly which code revision was verified.
Less guesswork. More confidence in the code we ship.
I want CSM to be more than a way to run multiple coding agents. I want it to help make the entire process of delegating engineering work more reliable.
github.com/onoja123/csm
Built CSM 🫡
A CLI for running multiple Claude Code, Codex, Gemini CLI & GitHub Copilot CLI accounts on one machine.
Switch accounts without logging out, keep your session, and automatically fail over to another account when one hits its usage limit.
Your tokens/cookies/Keychain secrets stay with the agents. CSM doesn’t store or copy them.
Open source → github.com/onoja123/csm
Akintade Oluwaseun retweeted
Build Receipts are now available in CSM Desktop and CLI, giving you a clear record of what actually ran, what passed, and what still needs attention.
The goal isn’t to promise bug-free software, but to make AI-generated work easier to inspect, verify, and trust.
Building in the open: github.com/onoja123/csm
