Menu
Restart local app/dev servers safely through the servers CLI/MCP using lifecycle locks, readiness waits, and operation traces.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Debug local app/dev server lifecycle problems with servers CLI/MCP status, logs, operations, traces, and lock evidence.
Start or register local app/dev servers through the servers CLI/MCP with locks, readiness checks, logs, and agent traces.
Based on SOC occupation classification
| name | skill-server-restart |
| description | Restart local app/dev servers safely through the servers CLI/MCP using lifecycle locks, readiness waits, and operation traces. |
| user_invocable | true |
Use this when an existing app/dev server must be restarted after code, env, dependency, or configuration changes.
mcp__servers__restart_local_server, mcp__servers__get_local_server_status, and mcp__servers__stop_local_server.servers servers:restart, servers servers:status, servers servers:debug, and servers servers:logs.pkill, kill -9, or restart by hand while a server-runtime lock exists. Inspect the lock and coordinate with the owner.--agent <name> and --reason <why>.servers servers:init ... --force first.Inspect the current registered state:
servers servers:status <slug> --refresh
servers servers:debug <slug>
If another agent owns the lifecycle lock, either wait intentionally or coordinate:
servers servers:restart <slug> --agent <agent> --reason '<reason>' --wait-lock --lock-timeout 300000
Restart and wait for readiness:
servers servers:restart <slug> --agent <agent> --reason '<reason>' --timeout 60000 --stop-timeout 15000
Confirm the new process is healthy:
servers servers:status <slug> --refresh
servers servers:logs <slug> --lines 80
If readiness fails, use servers servers:debug and logs. The lifecycle runtime should clean up a failed spawned process; do not start another unmanaged process to compensate.
The restart operation is completed, the observed status is online/ready, the PID/log metadata is current, and traces show who restarted the server and why.