One place to coordinate the work

I like having a choice of agents. Sometimes I want to talk through an idea. Sometimes I want a coding worker to take a specific task and return a change I can test. Sometimes the work is mostly research. There is no reason all of those jobs have to use the same model or interface.

What I don't want is to become the person manually carrying context between every tool. Explain the project here, paste the plan there, find the right session, ask whether it finished, then carry the answer back. At that point I'm doing a lot of coordination without getting much work done.

Keep the job connected

I use Hermes as the place to bring that work together. Auren helps with research and daily tasks, while coding work can go to agents such as Cursor and Codex. Cursor Forge is one of the projects I'm developing to make the Hermes and Cursor part of that process easier.

The point is to keep the job connected as it moves between tools. The worker should get the result I want, the context it needs and the limits on what it can change. When it comes back, I should be able to tell which task the output belongs to and what happened during the run.

That becomes especially useful when work needs another attempt. A review finding should lead to a focused followup on the same job, not a fresh conversation where we reconstruct what the first agent was trying to do.

More than starting workers

Launching several agents is only part of it. I also need to know when a task is stuck, whether the result has been checked and what still needs a decision from me. More workers don't help if the unfinished work becomes harder to follow.

I want to spend my attention on the questions that matter: is this the right thing to build, does the result solve the problem, and what should we do next?

That's the direction I'm working toward. I still want to choose the right tools for each job. I just want the plan, the work and the result to stay connected when I do.