Phase 1 - Planning: Create an implementation plan from the spec
- Architect (Opus): Create plan (direct mode, no interview)
- Critic (Opus): Validate plan
- Output:
.omc/plans/autopilot-impl.md
Phase 2 - Execution: Implement the plan using Ralph + Ultrawork
- Executor-low (Haiku): Simple tasks
- Executor (Sonnet): Standard tasks
- Executor-high (Opus): Complex tasks
- Run independent tasks in parallel
Phase 3 - QA: Cycle until all tests pass (UltraQA mode)
- Build, lint, test, fix failures
- Repeat up to 5 cycles
- Stop early if the same error repeats 3 times (indicates a fundamental issue)
Phase 4 - Validation: Multi-perspective review in parallel
- Architect: Functional completeness
- Security-reviewer: Vulnerability check
- Code-reviewer: Quality review
- All must approve; fix and re-validate on rejection
Phase 5 - Cleanup: Delete all state files on successful completion
- Remove
.omc/state/autopilot-state.json, ralph-state.json, ultrawork-state.json, ultraqa-state.json
- Run
/oh-my-claudecode:cancel for clean exit
Optional settings in .claude/settings.json:
{
"omc": {
"autopilot": {
"maxIterations": 10,
"maxQaCycles": 5,
"maxValidationRounds": 3,
"pauseAfterExpansion": false,
"pauseAfterPlanning": false,
"skipQa": false,
"skipValidation": false
}
}
}
Resume
If autopilot was cancelled or failed, run /oh-my-claudecode:autopilot again to resume from where it stopped.
Best Practices for Input
- Be specific about the domain -- "bookstore" not "store"
- Mention key features -- "with CRUD", "with authentication"
- Specify constraints -- "using TypeScript", "with PostgreSQL"
- Let it run -- avoid interrupting unless truly needed
Troubleshooting
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.
1---2name: autopilot-83description: Full autonomous execution from idea to working code4---5
6<Purpose>
7Autopilot takes a brief product idea and autonomously handles the full lifecycle: requirements analysis, technical design, planning, parallel implementation, QA cycling, and multi-perspective validation. It produces working, verified code from a 2-3 line description.
8</Purpose>
9
10<Use_When>
11- User wants end-to-end autonomous execution from an idea to working code
12- User says "autopilot", "auto pilot", "autonomous", "build me", "create me", "make me", "full auto", "handle it all", or "I want a/an..."
13- Task requires multiple phases: planning, coding, testing, and validation
14- User wants hands-off execution and is willing to let the system run to completion
15</Use_When>
16
17<Do_Not_Use_When>
18- User wants to explore options or brainstorm -- use `plan` skill instead
19- User says "just explain", "draft only", or "what would you suggest" -- respond conversationally
20- User wants a single focused code change -- use `ralph` or delegate to an executor agent
21- User wants to review or critique an existing plan -- use `plan --review`
22- Task is a quick fix or small bug -- use direct executor delegation
23</Do_Not_Use_When>
24
25<Why_This_Exists>
26Most 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.
27</Why_This_Exists>
28
29<Execution_Policy>
30- Each phase must complete before the next begins
31- Parallel execution is used within phases where possible (Phase 2 and Phase 4)
32- QA cycles repeat up to 5 times; if the same error persists 3 times, stop and report the fundamental issue
33- Validation requires approval from all reviewers; rejected items get fixed and re-validated
34- Cancel with `/oh-my-claudecode:cancel` at any time; progress is preserved for resume
35</Execution_Policy>
36
37<Steps>
381. **Phase 0 - Expansion**: Turn the user's idea into a detailed spec
39 - Analyst (Opus): Extract requirements
40 - Architect (Opus): Create technical specification
41 - Output: `.omc/autopilot/spec.md`
42
432. **Phase 1 - Planning**: Create an implementation plan from the spec
44 - Architect (Opus): Create plan (direct mode, no interview)
45 - Critic (Opus): Validate plan
46 - Output: `.omc/plans/autopilot-impl.md`
47
483. **Phase 2 - Execution**: Implement the plan using Ralph + Ultrawork
49 - Executor-low (Haiku): Simple tasks
50 - Executor (Sonnet): Standard tasks
51 - Executor-high (Opus): Complex tasks
52 - Run independent tasks in parallel
53
544. **Phase 3 - QA**: Cycle until all tests pass (UltraQA mode)
55 - Build, lint, test, fix failures
56 - Repeat up to 5 cycles
57 - Stop early if the same error repeats 3 times (indicates a fundamental issue)
58
595. **Phase 4 - Validation**: Multi-perspective review in parallel
60 - Architect: Functional completeness
61 - Security-reviewer: Vulnerability check
62 - Code-reviewer: Quality review
63 - All must approve; fix and re-validate on rejection
64
656. **Phase 5 - Cleanup**: Delete all state files on successful completion
66 - Remove `.omc/state/autopilot-state.json`, `ralph-state.json`, `ultrawork-state.json`, `ultraqa-state.json`
67 - Run `/oh-my-claudecode:cancel` for clean exit
68</Steps>
69
70<Tool_Usage>
71- Before first MCP tool use, call `ToolSearch("mcp")` to discover deferred MCP tools
72- Use `ask_codex` with `agent_role: "architect"` for Phase 4 architecture validation
73- Use `ask_codex` with `agent_role: "security-reviewer"` for Phase 4 security review
74- Use `ask_codex` with `agent_role: "code-reviewer"` for Phase 4 quality review
75- Agents form their own analysis first, then consult Codex for cross-validation
76- If ToolSearch finds no MCP tools or Codex is unavailable, proceed without it -- never block on external tools
77</Tool_Usage>
78
79<Examples>
80<Good>
81User: "autopilot A REST API for a bookstore inventory with CRUD operations using TypeScript"
82Why good: Specific domain (bookstore), clear features (CRUD), technology constraint (TypeScript). Autopilot has enough context to expand into a full spec.
83</Good>
84
85<Good>
86User: "build me a CLI tool that tracks daily habits with streak counting"
87Why good: Clear product concept with a specific feature. The "build me" trigger activates autopilot.
88</Good>
89
90<Bad>
91User: "fix the bug in the login page"
92Why bad: This is a single focused fix, not a multi-phase project. Use direct executor delegation or ralph instead.
93</Bad>
94
95<Bad>
96User: "what are some good approaches for adding caching?"
97Why bad: This is an exploration/brainstorming request. Respond conversationally or use the plan skill.
98</Bad>
99</Examples>
100
101<Escalation_And_Stop_Conditions>
102- Stop and report when the same QA error persists across 3 cycles (fundamental issue requiring human input)
103- Stop and report when validation keeps failing after 3 re-validation rounds
104- Stop when the user says "stop", "cancel", or "abort"
105- If requirements were too vague and expansion produces an unclear spec, pause and ask the user for clarification before proceeding
106</Escalation_And_Stop_Conditions>
107
108<Final_Checklist>
109- [ ] All 5 phases completed (Expansion, Planning, Execution, QA, Validation)
110- [ ] All validators approved in Phase 4
111- [ ] Tests pass (verified with fresh test run output)
112- [ ] Build succeeds (verified with fresh build output)
113- [ ] State files cleaned up
114- [ ] User informed of completion with summary of what was built
115</Final_Checklist>
116
117<Advanced>
118## Configuration
119
120Optional settings in `.claude/settings.json`:
121
122```json
123{
124 "omc": {
125 "autopilot": {
126 "maxIterations": 10,
127 "maxQaCycles": 5,
128 "maxValidationRounds": 3,
129 "pauseAfterExpansion": false,
130 "pauseAfterPlanning": false,
131 "skipQa": false,
132 "skipValidation": false
133 }
134 }
135}
136```
137
138## Resume
139
140If autopilot was cancelled or failed, run `/oh-my-claudecode:autopilot` again to resume from where it stopped.
141
142## Best Practices for Input
143
1441. Be specific about the domain -- "bookstore" not "store"
1452. Mention key features -- "with CRUD", "with authentication"
1463. Specify constraints -- "using TypeScript", "with PostgreSQL"
1474. Let it run -- avoid interrupting unless truly needed
148
149## Troubleshooting
150
151**Stuck in a phase?** Check TODO list for blocked tasks, review `.omc/autopilot-state.json`, or cancel and resume.
152
153**QA cycles exhausted?** The same error 3 times indicates a fundamental issue. Review the error pattern; manual intervention may be needed.
154
155**Validation keeps failing?** Review the specific issues. Requirements may have been too vague -- cancel and provide more detail.
156</Advanced>
majiayu000