Skip to main content
subagent CX Codex

agents_list — Claude Code subagent: what it does, how to install

agents_list — a subagent for Claude Code and Codex monitoring running herdr sessions across user machines. Displays active jobs, pending approvals, and subscription limits.

MIT ★ 395 updated 2026-11-16

Install into Claude Code · Manually: put the files in place All methods and platforms ↓
.claude/agents/agents_list.md
Open on GitHub

What agents_list does

This subagent queries herdr to inspect Claude Code and Codex processes running on connected machines. It pulls session identifiers, target machines, active project directories, and the last emitted response. By default, the active filter bundles tasks flagged as needs_you, working, and unread.

Output lines include pending user questions with interaction_id references, launcher targets, and usage_limit flags with reset timestamps. For Claude Code, it indicates ultracode=on/off state. The summary also lists launchers that failed their latest health check.

What it helps with

  • track agent sessions paused on user approval
  • filter running processes by machine, project, or state
  • spot failing launchers across connected computers
  • detect subscription usage limits across active agents

Example prompts for the agent

  • Show active agents waiting for approval
  • List herdr agents on machine devbox for persona project
  • Get finished agents from the last 24 hours

How to install agents_list

Manually: put the files in place

.claude/agents/agents_list.md

For all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention

Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean

Questions about agents_list

How to install agents_list in Claude Code?

Add the subagent definition file directly to .claude/agents/agents_list.md.

What does agents_list do?

It monitors herdr sessions across computers, reporting process state, pending questions, launcher health, and usage limits.

What is required to run agents_list?

A working herdr setup across your machines running Claude Code or Codex.

Source text

agents_list.md · MIT
The Claude Code / Codex agents running on the user's computers (herdr), with their state: needs_you (waiting for an answer / approval), working, unread (finished, result not looked at), idle. done lists work finished in the last 24 hours, and muted lists the muted agents that would otherwise appear in the live snapshot; both include the work conclusion when known. Each line has machine, session id, project, title, last update and the last thing the agent said. launcher= names the launcher an agent runs on when it is known, and usage_limit (with resets_at when the reset is known) marks one its subscription's usage limit stopped (agent_switch moves such an agent to another launcher when the user asks). Known Claude Code switch states appear as ultracode=on/off; an omitted state is unknown. ultracode=on means every substantive task runs a multi-agent dynamic workflow, independently of thinking effort. Let it finish and wait for workflow results without nudging it. Much faster and more accurate than ssh + herdr. At the end: the launchers that failed their last health check, by machine (agent_start and task_plan move off them).

A line under an agent is something waiting for the user: a question lists its options and the interaction id to answer it with (interaction_answer), an approval is the user's to give in the app.

```json
{
  "type": "object",
  "properties": {
    "state": {
      "type": "string",
      "enum": ["active", "needs_you", "working", "unread", "idle", "done", "muted", "archived", "all"],
      "description": "default active = needs_you + working + unread; done = finished work in the last 24 hours; muted = muted agents in the live snapshot categories; archived: only the archived sessions (agent_unarchive restores them); the others leave archived ones out, all lists them marked archived"
    },
    "machine": {
      "type": "string"
    },
    "query": {
      "type": "string",
      "description": "filter by title / project / directory"
    }
  }
}
```

© nwverse · MIT · nwverse/persona

Similar subagents

All subagents for Agent setup: memory, context, orchestration →

From the same set61

The whole set →
agent_autoname_receipt The agent_autoname_receipt Claude Code subagent reads existing name-confirmation cards and one coding agent’s current naming snapshot. It helps inspect the available evidence about a rename. CX Codex ★ 395repo stars ask_escalate The ask_escalate Claude Code subagent routes decisions that task agents cannot make to the user. It groups unanswered task questions into one reminder. CX Codex ★ 395repo stars agent_supervise The agent_supervise Claude Code subagent takes over an agent that was started without a task and tracks its work as a task in the current conversation. It routes progress messages to a supervisor and closes the task with task_review. CX Codex ★ 395repo stars task_decided The task_decided Claude Code subagent records an owner's answer to a pending task decision. It captures the choice when the owner gives it in the conversation. CX Codex ★ 395repo stars task_scope task_scope — a subagent for Claude Code revises a supervised task after it starts. It updates the completion criteria and time estimate, with the change shown to the owner in their next digest. CX Codex ★ 395repo stars agent_archive Subagent for Claude Code moves a coding agent session from the active list to the archive. It triggers when cleaning up or archiving agent sessions. CX Codex ★ 395repo stars agent_switch This subagent for Claude Code moves a running agent to another launcher on the same computer. It switches sessions only when the user asks. CX Codex ★ 395repo stars worker_supervise worker_supervise — a subagent for Claude Code supervises background workers launched with task_plan. It checks submitted evidence against completion criteria and updates worker status via task_review. CX Codex ★ 395repo stars

Project sponsors

Companies that keep this analytics open