mohit17mor

SquadAI is the infrastructure layer for running Codex agents as part of your systems not just as individual chat windows.

It lets your codex agents wake up and react to changes (events) in the world. Example of changes/events :

Ops tickets coming in, your script/monitor detects it and sends over, New work item created in jira, Flight price changes, Or anything around which a script/monitor can be written which can then send events to SquadAI

Not only that, SquadAI also handles routing it to the right agent, queueing it, creating or resuming agent instances, managing persistent threads, shared library of skills, coordinating work across machines and giving you one place to see and control it all.

Adding and connecting multiple machine can be easily done using the tailscale support. I wanted it to be super easy to add any machine where the projects are already setup.

Also, added the support to talk to multiple agents through telegram group, the agents could be on any machine and still have the relevant context of the group chat!

It internally uses codex-app-server to manage the codex threads. Added a wrapper code around the app-server apis to handle the lifecycle of the threads.

This is super cool! Had similar idea but couldn't pull it off, kudos to the team! Few questions: 1. Is the long term aim to build 1 super app kind of thing? Like for example if i want to book a cab, order food, search the web etc, this app will take care of it all?

2. Any plans to make it like a service? Maybe people could integrate it with their apps/websites?

I am exploring the idea of event driven agents to reduce human inputs as much as possible, basically I should be able to just create tasks and the agents just keep picking them up as they get free. Also, trying to extend it to chaining the agents so that it can create an end to end workflow. For coding it might look something like this: User creates a task -> coder picks it up -> reviewer agent -> (coder/reviewer back and forth upto 4-5 turns) -> tester agent -> finally wait for user to merge the code or not.

This has nothing to do with agent harness as such, its more like treating agents as instantiable objects and adding control plane on top of them to get the work done.

Not really sure if you mean setups like OpenClaw as well, but I ran into pretty similar issues there.

That was actually a big part of why I started building my own agent system, Arc agent, mostly just to see if I could solve some of the problems every agent faces. A few practical things helped, putting limits on iteration/retry loops so the agent can’t wander forever, being stricter about context/tool handling once external skills are in play and adding a simple cost/token counter so I could at least see where usage was going. I also tried to reduce the usage of llm whereever a simple code would work more reliably.

I tried to tackle few other problems like downtime during context compaction etc, but yeah the project is still work in progress and i am experimenting with diff stuff.