Agent Building skills
Browse reusable Agent Skills, each with a clear purpose and practical guidance.
goal-griller
Use when the user wants to turn a rough idea, vague task, feature wish, bug-fix intent, or Codex /goal objective into a clear verifiable goal prompt. Use when the user mentions goal mode, /goal, set_goal, long-running autonomous Codex work, or asks to be interviewed/grilled before creating a goal.
launching-agent-teams
Launches agent teams with structured roles and task decomposition. Use when asked to create a team, spawn teammates, or coordinate multiple agents in parallel.
ai-error-states
Use when implementing handling AI-specific errors.
arksim-evaluate
Use when the user wants to re-evaluate a previous arksim simulation with different metrics, thresholds, or judge model without re-running the agent. Cheaper than re-simulating.
arksim-results
Use when the user wants to inspect arksim evaluation results, debug specific failures turn by turn, or compare two runs to measure improvement.
arksim-simulate
Use when the user wants to simulate multi-turn conversations against an AI agent. Alias for the arksim-test skill; the canonical flow lives there.
arksim-test
Use when the user wants to test, simulate, or evaluate an AI agent against multi-turn scenarios (also exposed as the arksim-simulate alias). Discovers the agent, generates scenarios, runs simulation and evaluation, surfaces failures.
claude-code-narp-runtime
当用户希望把 Claude Code 作为 NextClaw 正式会话类型接入，尤其是通过 narp-stdio 配置、安装、修复、doctor、补齐 Runtime default 模型选项或真实冒烟时使用。
clawhub-nextclaw
Use when the user wants to search external ClawHub skills or install a ClawHub skill into a NextClaw workspace's skills directory without treating ClawHub as the NextClaw marketplace.