Conversation monitor — a second Claude agent that watches the front agent's work from outside and nudges when trajectory goes wrong. Reads session JSONL incrementally, maintains a compact judgment trace, delivers nudges via hook.
$npx -y skills add horiacristescu/claude-playbook-plugin --skill monitorInstalls into the current project.
Run `npx skills use "https://github.com/horiacristescu/claude-playbook-plugin" --skill "horiacristescu/claude-playbook-plugin/monitor"` and follow the generated skill instructions now. Read its complete output, redirecting it to a temporary file first if necessary. Resolve relative paths from the supporting-files directory it provides.
Use the skills in "https://github.com/horiacristescu/claude-playbook-plugin" that are relevant to the current task. Run `npx skills add "https://github.com/horiacristescu/claude-playbook-plugin"` and select the relevant skills, then follow their instructions.