How do I see what all my coding agents are doing?

Each major coding agent now has its own overview screen. In Claude Code, run claude agents to open agent view, one list of every background session with its state and which ones need you. In Codex CLI, codex agents opens the agents dashboard for your local tasks. In Cursor, the Agents Window shows its local and cloud agents across your repositories. Each one shows only its own agent's work. If you run more than one kind of agent, or your teammates run theirs, you need a view above all of them.

The question behind "what are my agents doing?" is usually narrower: which one is stuck waiting for me, which one finished, and which two are about to touch the same thing. A good overview answers those three at a glance and lets you ignore the rest.

What does each agent show?#

They are good at what they cover. Agent view, for example, lets you dispatch a task as a new row, peek at a row to reply without opening the whole conversation, and attach when you want the full session. Background sessions keep running with no terminal attached.

How do I track agents by hand?#

Before a tool does it for you, three habits cover most of it:

1. Name every session after its task. "Fix flaky login test" in a list is readable; five sessions called claude are not. Claude Code's /rename and Codex's dashboard both let you rename.
2. Give every agent its own worktree and branch. git worktree list and git branch then tell you what is in flight, and nothing is half-edited in a shared folder.
3. Keep one task list. A short TASKS.md with who holds each task shows the work even when the terminals are closed.

Where do the built-in views stop?#

• One vendor each. Agent view shows Claude Code sessions; a Codex task beside it is invisible there, and the reverse.
• One computer each, mostly. A Claude Code or Codex session on your desktop at home does not appear on your laptop; Cursor's view reaches only Cursor's own cloud agents.
• One person each. Nothing shows what your teammate's agents hold, so two people's agents can start the same task.
• At the keyboard. When you step away, the only signal is whatever notification you set up.

Where does Joxo fit?#

Joxo shows a team's agent work in one place, whichever agent each person runs: Claude Code, Codex, Cursor, OpenCode and more. On the desktop, the phone and the web, Sessions lists your own agent sessions on each of your computers; you can watch one live, answer its questions and start a new one. The board shows which agent holds every task across the team, and only one agent can hold a task. When one of your agents needs you, the iPhone app (in TestFlight) tells you. Setup is one line pasted into each agent:

Read https://joxo.ai/skill.md and follow it to set up Joxo in this project.

Running several Claude Code sessions in parallel covers how to split the work in the first place.

Frequently asked questions#

Does agent view show sessions I started in a normal terminal?#

Agent view is built around background sessions: ones you dispatch from it, or start with --bg. Anthropic's agent view docs describe which sessions appear and how to move one into the background.

Can I see my agents from my phone?#

Partly, per agent. Claude Code's Remote Control opens a session in the Claude app, and some agents have their own mobile apps. Joxo's iPhone app shows your sessions on every computer, whatever the agent, and brings their questions to you.

How do I know which agent is waiting for me?#

Look for the state each overview shows: "needs input" in agent view, waiting tasks in the Codex dashboard. To be told rather than to look, set up notifications: getting a notification when Claude Code is waiting shows how.

Can I see what my teammates' agents are doing?#

Not from the built-in views, which show your own sessions. A shared task board does it: each agent claims its task there, so everyone sees who holds what. Joxo's board works that way across agents and computers.