autopilot

仓库创建 2026年7月22日最近提交 3 天前SkillHot 收录 4 小时前

▸ 精选理由

适合需要端到端自动化与多轮审查的复杂任务执行场景。

▸ 风险提示

可能调用子进程/执行 shell 命令，具较高操作风险

这个 Skill 做什么

以编排者身份拆分任务、调度子代理执行并复核结果直到合格的全自动工作流。

把一项复杂任务分解成子任务，派子代理去执行、再由编排器（默认用 Opus，必要时用 Fable）复核，直到结果合格为止，用户不需要逐步指挥。适合想要“交给我全自动跑完”的多步工作或端到端交付。特点是自驱式调度、分工执行并有复核和修正回路（默认修正两轮），能处理复杂、多阶段的流程。

▸ 展开 SKILL.md 英文原文

Fully self-driven task execution — Opus/Fable plans the work into a task list, subagents execute each task (escalating to Opus/Fable when a task is judged complex), then Opus/Fable reviews everything for bugs/improvements and loops fixes back until clean. Use when the user says "autopilot", "/autopilot <task>", "full auto", "run this end to end", or wants a multi-step task self-orchestrated without step-by-step guidance from them.

Skill 开发管理任务编排子代理自动执行通用

Stars

Forks

仓库内 Skill

积累中

7 日增星

安装 / 使用

给你的 Agent 一句话（通用）

帮我安装这个 skill：https://raw.githubusercontent.com/thititongumpun/skills/main/skills/autopilot/SKILL.md

或 curl 直取 SKILL.md

curl -fsSL "https://raw.githubusercontent.com/thititongumpun/skills/main/skills/autopilot/SKILL.md"

SKILL.MD 节选查看完整文件 ↗

# Autopilot

You are an ORCHESTRATOR. Do not do the work yourself — deploy subagents for
planning, execution, and review, and coordinate between them.

Default planner/reviewer model is `opus`. Use `fable` instead only if the
user asked for it. Default fix-loop cap is 2 rounds unless the user says
otherwise.

If `rtk` is on the machine, shell commands — yours and every subagent's —
go through it (`rtk git status`, `rtk cargo test`): same output, far fewer
tokens. Say so in each subagent's brief; they don't inherit this.

## Progress reporting

The user is not driving this loop, so tell them where it is without being
asked.

- After Phase 1, print the full numbered task list, marking which ta

via SKILL·HOT · 数据来自 GitHub 公开信息 · 原文版权归作者所有