SniporaSee the dev servers Claude Code and Codex left running
Coding agents start dev servers to check their work, and don’t always stop them. Three sessions later, you have six servers and no idea which are yours.
In short
Portlivo looks up each server’s process tree. When Claude Code or Codex is in it, the row gets the agent’s logo and the Agents filter shows only those, ready to stop with Stop Visible. Nothing to install in the agent.
How it knows
Every scan reads the process table once. If the server, or anything above it, is Claude Code or Codex (including when they run under node), that’s the agent. The app you started the agent from, such as iTerm2 or Cursor, is still shown as where it came from.
Clearing up after a session
- Open the panel and choose the Agents filter.
- Check the list: each row shows its project folder and how long it’s been up.
- Press Stop Visible and confirm once. Containers and services are left alone.
Typing claude or codex in the search box works too.
Questions people ask
- Does it work with other agents?
- Today it recognises Claude Code and Codex. Servers from other tools still show up like any other, with their project and the app they came from.
- Does it read what the agent is doing?
- No. It only looks at process names and command lines, the same as Activity Monitor.
Try Portlivo free for 7 days. Every feature, no account.
Download for MacSend to my MacStill have a question?
Write to us — a real developer reads every email and replies personally.