Skip to main content
subagent CX Codex

launcher_repair — Claude Code subagent: features and setup

launcher_repair — a subagent for Claude Code and Codex repairs agent CLI launchers on a specified machine. It tests each launcher after making changes.

MIT ★ 395 updated 2026-11-16

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

What launcher_repair does

The subagent handles launcher failures detected by the hub’s health check and repairs agent CLI setup on the target machine. For a model routing error, it resets the gateway’s model names; for an authentication failure, it configures the gateway key again. It upgrades an outdated CLI, while investigating timeouts and other errors first.

Its work stays within the machine’s agent CLI setup. It leaves projects untouched, and does not push, deploy, or message anyone. After repairs, it runs each launcher non-interactively and reports whether it works or why it remains broken. The hub checks the launchers again after the subagent stops.

What it helps with

  • reset gateway model names rejected by a launcher
  • restore access after a gateway key is missing or rejected
  • upgrade an outdated agent CLI
  • investigate a launcher timeout before attempting a repair

Example prompts for the agent

  • Repair the launchers that the hub health check marked as broken on this machine.
  • Fix the gateway authentication error in the agent CLI setup on this machine.
  • Investigate why a launcher fails, then test it with a non-interactive run.

How to install launcher_repair

Manually: put the files in place

.claude/agents/launcher_repair.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 launcher_repair

How do I install launcher_repair in Claude Code?

Add the subagent file at .claude/agents/launcher_repair.md.

What does launcher_repair do?

It repairs agent CLI setup on a machine where the hub found broken launchers, then tests each launcher.

Which launcher failures can launcher_repair handle?

It can reset gateway model names, configure a gateway key again, and upgrade an outdated CLI. It investigates timeouts and other errors before acting.

Source text

launcher_repair.md · MIT
The hub's health check found launchers that do not work on the machine {{.Machine}}{{if .Name}} ({{.Name}}){{end}}{{if .SSHHost}}, reachable as `ssh {{.SSHHost}}`{{end}}. Repair them with the {{.Skill}} skill (it installs and configures Claude Code and Codex and connects them to their gateway).

{{range .Launchers}}- `{{.Name}}` ({{.Agent}}), {{.Kind}}: {{.Error}}
{{end}}
What the classes usually mean: model_route: the gateway does not accept the model name the launcher sends (set the gateway's model names again); auth: the gateway key is missing, expired or rejected (configure the key again); cli_outdated: the CLI is too old (upgrade it); timeout / other: find out why first.

Work only on that machine's agent CLI setup: don't touch its projects, and don't push, deploy or message anyone. When you are done, run each launcher once there non-interactively (for Claude Code: `<launcher> -p "Reply with exactly one word: ok" --max-turns 1`) and finish with one line per launcher: fixed, or why it is still broken. The hub checks them again after you stop.

© nwverse · MIT · nwverse/persona

Similar subagents

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

From the same set61

The whole set →
direction_name The direction_name Claude Code subagent assigns a name to a direction linked to a leader conversation. The app and work-log extension display that label. CX Codex ★ 395repo stars question_moved_to_todo The question_moved_to_todo Claude Code subagent moves an unanswered question into the user's to-do list. It waits for a decision while the agent continues with work that does not depend on it. CX Codex ★ 395repo stars task_plan_withdraw The task_plan_withdraw Claude Code subagent withdraws a plan that has not been started. It runs when a plan from the current conversation is no longer needed. CX Codex ★ 395repo stars question_asked_later The question_asked_later Claude Code subagent sends a question to the user as a card on their phone. It finishes unrelated work and ends the turn without guessing while the answer is pending. CX Codex ★ 395repo stars task_spawn The task_spawn Claude Code subagent starts a separate agent for part of an active assignment. The new agent joins the same to-do list and reports to the same supervisor. CX Codex ★ 395repo stars task_leader The task_leader subagent for Claude Code splits work among child agents and checks their results against defined completion criteria. It handles review decisions for each delegated task. CX Codex ★ 395repo stars task_start_keys The task_start_keys Claude Code subagent types text and presses keys in a task’s terminal tab after its agent stops there. It can answer a shell prompt or move past the screen’s current point. CX Codex ★ 395repo stars agent_autoname The agent_autoname Claude Code subagent renames an agent session using its conversation. It runs only after an explicit Auto Rename request. CX Codex ★ 395repo stars

Project sponsors

Companies that keep this analytics open