Agent skill

autopilot

全自动 RPI 执行 — 从想法到可运行代码

Stars 163
Forks 31

Install this agent skill to your Project

npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/autopilot-1molchuan-oh-my-ccg

SKILL.md

oh-my-ccg Autopilot

Full automatic RPI cycle: research → plan → impl → review.

Activation

Trigger: "autopilot", "auto" + requirement description

Workflow

Phase 1: Initialize

  • Use rpi_state_write MCP tool to create RPI state with requirement
  • Verify Codex + Gemini availability via ask_codex(prompt="ping") and ask_gemini(prompt="ping") MCP calls
  • Use mode_state_write MCP tool to start autopilot state tracking

Phase 2: Auto-Research

  • Run explore agent for codebase context
  • Launch multi-model exploration in parallel via MCP tools:
    • ask_codex(agent_role="analyst", background=true) — backend constraints
    • ask_gemini(agent_role="designer", background=true) — frontend constraints
  • Collect results via check_job_status
  • Synthesize constraints automatically
  • Present constraint summary to user for quick confirmation

Phase 3: Auto-Plan

  • Suggest /clear if context > 80%
  • Run planner agent with constraints
  • Cross-validate with dual-model parallel MCP calls:
    • ask_codex(agent_role="planner", background=true) — plan validation
    • ask_gemini(agent_role="designer", background=true) — frontend validation
  • Generate zero-decision task list
  • Update state via rpi_state_write MCP tool

Phase 4: Auto-Impl

  • Suggest /clear if context > 80%
  • Route tasks via MCP tools:
    • Frontend tasks → ask_gemini for prototype, then executor rewrites
    • Backend tasks → ask_codex for prototype, then executor rewrites
  • Use Team mode if multiple independent tasks
  • Use Ralph mode for each task (execute→verify→fix)
  • Continue until all tasks pass verification
  • Update state via rpi_state_write MCP tool after each task

Phase 5: Auto-Review

  • Run dual-model cross-review via parallel MCP calls:
    • ask_codex(agent_role="code-reviewer", background=true)
    • ask_gemini(agent_role="designer", background=true)
  • Fix Critical findings automatically
  • Present final report to user

Context Management

  • Monitor context usage after each phase
  • Prompt user for /clear when approaching 80%
  • State persists in .oh-my-ccg/state/ across /clear (managed via MCP tools)

Cancellation

Use /oh-my-ccg:cancel to stop autopilot at any point.

Expand your agent's capabilities with these related and highly-rated skills.

Didn't find tool you were looking for?

Be as detailed as possible for better results