Claude Sessions — running many AI agents at once

Context: personal project · 2026 Stack: Claude Code, Python, desktop app

The problem

When several AI coding agents work in parallel — each on its own task, some spawning sub-agents — it gets hard to see which one is stuck, which is waiting for you, and how much each is spending.

What I built

A desktop app for managing many Claude Code agents from one place:

• Status alerts — know at once when an agent finishes, fails or needs a decision.
• Sub-agent orchestration — see which agent started which sub-agent, and pass messages between them.
• Context and token management — keep each agent's context and spend visible and under control.

What it shows

I work with AI agents daily, at the level of how they are orchestrated and kept within budget — not just by prompting them.