autopilot by yeachan-heo/oh-my-claudecode
npx skills add https://github.com/yeachan-heo/oh-my-claudecode --skill autopilot<使用时机>
<请勿使用时机>
plan 技能ralph 或委托给执行器代理plan --review<存在原因> 大多数非平凡的软件任务需要协调的阶段:理解需求、设计解决方案、并行实现、测试和验证质量。自动驾驶自动编排所有这些阶段,使用户可以描述他们想要什么,并在无需管理每个步骤的情况下获得可运行的代码。 </存在原因>
<执行策略>
/oh-my-claudecode:cancel 取消;进度将被保存以便恢复 </执行策略>第 1 阶段 - 规划:根据规范创建实施计划
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
.omc/plans/autopilot-impl.md第 2 阶段 - 执行:使用 Ralph + Ultrawork 实施计划
第 3 阶段 - QA:循环直到所有测试通过(UltraQA 模式)
第 4 阶段 - 验证:并行进行多角度审查
第 5 阶段 - 清理:成功完成后删除所有状态文件
.omc/state/autopilot-state.json、ralph-state.json、ultrawork-state.json、ultraqa-state.json/oh-my-claudecode:cancel 以干净退出<工具使用>
Task(subagent_type="oh-my-claudecode:architect", ...) 进行第 4 阶段的架构验证Task(subagent_type="oh-my-claudecode:security-reviewer", ...) 进行第 4 阶段的安全审查Task(subagent_type="oh-my-claudecode:code-reviewer", ...) 进行第 4 阶段的质量审查<升级与停止条件>
/deep-interview 进行苏格拉底式澄清,或在继续之前暂停并请用户澄清 </升级与停止条件><最终检查清单>
.claude/settings.json 中的可选设置:
{
"omc": {
"autopilot": {
"maxIterations": 10,
"maxQaCycles": 5,
"maxValidationRounds": 3,
"pauseAfterExpansion": false,
"pauseAfterPlanning": false,
"skipQa": false,
"skipValidation": false
}
}
}
如果自动驾驶被取消或失败,请再次运行 /oh-my-claudecode:autopilot 以从中断处恢复。
卡在某个阶段? 检查 TODO 列表中的阻塞任务,查看 .omc/autopilot-state.json,或取消并恢复。
QA 循环已用尽? 同一错误出现 3 次表明存在根本问题。请检查错误模式;可能需要手动干预。
验证持续失败? 检查具体问题。需求可能过于模糊——请取消并提供更多细节。
当使用模糊输入调用自动驾驶时,第 0 阶段可以重定向到 /deep-interview 进行苏格拉底式澄清:
用户:"autopilot build me something cool"
自动驾驶:"您的请求是开放式的。您想先进行一次深度访谈吗?"
[是的,先进行访谈(推荐)] [不,直接扩展]
如果深度访谈规范已存在于 .omc/specs/deep-interview-*.md,自动驾驶将直接使用它作为第 0 阶段的输出(该规范已经过数学验证以确保清晰度)。
推荐的完整流水线链接了三个质量关卡:
/deep-interview "模糊想法"
→ 苏格拉底式问答 → 规范(模糊度 ≤ 20%)
→ /ralplan --direct → 共识计划(规划师/架构师/批评家批准)
→ /autopilot → 跳过第 0+1 阶段,从第 2 阶段(执行)开始
当自动驾驶检测到 ralplan 共识计划(.omc/plans/ralplan-*.md 或 .omc/plans/consensus-*.md)时,它会跳过第 0 阶段(扩展)和第 1 阶段(规划),因为该计划已经过:
自动驾驶直接从第 2 阶段开始(通过 Ralph + Ultrawork 执行)。
每周安装次数
165
仓库
GitHub 星标数
11.2K
首次出现
2026 年 1 月 22 日
安全审计
安装于
opencode152
gemini-cli146
cursor145
codex144
claude-code142
github-copilot135
<Use_When>
<Do_Not_Use_When>
plan skill insteadralph or delegate to an executor agentplan --review<Why_This_Exists> Most non-trivial software tasks require coordinated phases: understanding requirements, designing a solution, implementing in parallel, testing, and validating quality. Autopilot orchestrates all of these phases automatically so the user can describe what they want and receive working code without managing each step. </Why_This_Exists>
<Execution_Policy>
/oh-my-claudecode:cancel at any time; progress is preserved for resume </Execution_Policy>Phase 1 - Planning : Create an implementation plan from the spec
.omc/plans/autopilot-impl.mdPhase 2 - Execution : Implement the plan using Ralph + Ultrawork
Phase 3 - QA : Cycle until all tests pass (UltraQA mode)
Phase 4 - Validation : Multi-perspective review in parallel
Phase 5 - Cleanup : Delete all state files on successful completion
.omc/state/autopilot-state.json, , , <Tool_Usage>
Task(subagent_type="oh-my-claudecode:architect", ...) for Phase 4 architecture validationTask(subagent_type="oh-my-claudecode:security-reviewer", ...) for Phase 4 security reviewTask(subagent_type="oh-my-claudecode:code-reviewer", ...) for Phase 4 quality review<Escalation_And_Stop_Conditions>
/deep-interview for Socratic clarification, or pause and ask the user for clarification before proceeding </Escalation_And_Stop_Conditions><Final_Checklist>
Optional settings in .claude/settings.json:
{
"omc": {
"autopilot": {
"maxIterations": 10,
"maxQaCycles": 5,
"maxValidationRounds": 3,
"pauseAfterExpansion": false,
"pauseAfterPlanning": false,
"skipQa": false,
"skipValidation": false
}
}
}
If autopilot was cancelled or failed, run /oh-my-claudecode:autopilot again to resume from where it stopped.
Stuck in a phase? Check TODO list for blocked tasks, review .omc/autopilot-state.json, or cancel and resume.
QA cycles exhausted? The same error 3 times indicates a fundamental issue. Review the error pattern; manual intervention may be needed.
Validation keeps failing? Review the specific issues. Requirements may have been too vague -- cancel and provide more detail.
When autopilot is invoked with a vague input, Phase 0 can redirect to /deep-interview for Socratic clarification:
User: "autopilot build me something cool"
Autopilot: "Your request is open-ended. Would you like to run a deep interview first?"
[Yes, interview first (Recommended)] [No, expand directly]
If a deep-interview spec already exists at .omc/specs/deep-interview-*.md, autopilot uses it directly as Phase 0 output (the spec has already been mathematically validated for clarity).
The recommended full pipeline chains three quality gates:
/deep-interview "vague idea"
→ Socratic Q&A → spec (ambiguity ≤ 20%)
→ /ralplan --direct → consensus plan (Planner/Architect/Critic approved)
→ /autopilot → skips Phase 0+1, starts at Phase 2 (Execution)
When autopilot detects a ralplan consensus plan (.omc/plans/ralplan-*.md or .omc/plans/consensus-*.md), it skips both Phase 0 (Expansion) and Phase 1 (Planning) because the plan has already been:
Autopilot starts directly at Phase 2 (Execution via Ralph + Ultrawork).
Weekly Installs
165
Repository
GitHub Stars
11.2K
First Seen
Jan 22, 2026
Security Audits
Gen Agent Trust HubPassSocketPassSnykPass
Installed on
opencode152
gemini-cli146
cursor145
codex144
claude-code142
github-copilot135
AI Elements:基于shadcn/ui的AI原生应用组件库,快速构建对话界面
63,800 周安装
Flutter布局指南:构建响应式UI的约束规则与自适应设计模式
1,200 周安装
Skill Judge:AI Agent Skill评估工具,官方规范与17+示例提炼,优化知识增量与思维模式
163 周安装
Dev Browser:JavaScript沙盒化浏览器控制CLI工具 - 自动化测试与爬虫开发利器
1,200 周安装
CTF Web漏洞利用速查手册 - SQL注入、SSTI、XXE、反序列化、XSS等攻防技术大全
1,300 周安装
OKX Onchain OS 审计日志技能 - 离线问题排查与日志文件路径指南
1,200 周安装
自动设置预提交钩子:Husky + lint-staged + Prettier 一键配置代码质量检查
1,300 周安装
ralph-state.jsonultrawork-state.jsonultraqa-state.json/oh-my-claudecode:cancel for clean exit