External Orchestrator with Worker-Judge Separation — spawns fresh claude -p sessions per iteration with adversarial two-skeptic verification. Arithmetic exit decisions only. Requires CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1.
$npx -y skills add kasempiternal/claude-agent-system --skill siegeInstalls into the current project.
Run `npx skills use "https://github.com/kasempiternal/claude-agent-system" --skill "kasempiternal/claude-agent-system/siege"` 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/kasempiternal/claude-agent-system" that are relevant to the current task. Run `npx skills add "https://github.com/kasempiternal/claude-agent-system"` and select the relevant skills, then follow their instructions.