菜单
Use when task intent matches command 'execute-ralph-cc'. Do not use for unrelated workflows.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use to audit test quality with Google Fellow SRE scrutiny - identifies tautological tests, coverage gaming, weak assertions, missing corner cases. Creates bd epic with tasks for improvements, then runs SRE task refinement on each.
Use when creating or developing anything, before writing code - refines rough ideas into bd epics with immutable requirements
Use when creating Claude Code hooks - covers hook patterns, composition, testing, progressive enhancement from simple to advanced
Use when encountering bugs or test failures - systematic debugging using debuggers, internet research, and agents to find root cause before fixing
Use when facing 3+ independent failures that can be investigated without shared state or dependencies - dispatches multiple agents to investigate and fix independent problems concurrently
Execute entire bd epic autonomously via subagent-per-task dispatch loop. Setup, dispatch subagent per task, end-of-epic review, branch completion.
| name | codex-command-execute-ralph-cc |
| description | Use when task intent matches command 'execute-ralph-cc'. Do not use for unrelated workflows. |
This skill wraps the source file commands/execute-ralph-cc.md for Codex Skills compatibility.
commands/execute-ralph-cc.md.---
name: execute-ralph-cc
description: "Execute entire bd epic autonomously via ScheduleWakeup task-per-turn loop. Claude Code only. No stop hooks."
type: flow
---
# Usage
```text
/xpowers:execute-ralph-cc [--reviewer-model=opus|sonnet]
```
## Arguments
- `--reviewer-model`: Model for autonomous-reviewer (optional)
- `opus` (default): Highest capability, thorough review
- `sonnet`: Faster, balanced quality
## What This Does
Executes a complete bd epic autonomously using Claude Code's native `ScheduleWakeup` tool instead of stop hooks. Each turn processes exactly one task, then schedules a wake-up for the next:
1. **Phase 0 - State Recovery:** Every wake-up starts here. Reads all state from bd/tm (epic, tasks, criteria). Fully idempotent -- any wake-up reconstructs context from bd/tm alone.
2. **Phase 1 - Get Task & Refine:** Uses `bv --robot-triage` output from Phase 0 to determine next task. Claims or auto-creates the next task. Runs SRE refinement.
3. **Phase 2 - Dispatch Subagent:** Records `PRE_SHA`, dispatches Agent tool with subagent-driven-development protocol. Verifies SHA drift and task status.
4. **Phase 3 - Post-Task Check:** Quick review. Re-reads epic criteria. If criteria unmet, calls `ScheduleWakeup(60s)` and ends turn. If criteria met, proceeds to Phase 4.
5. **Phase 4 - End-of-Epic Review:** Dispatches 3 specialized agents in parallel (review-quality, security-scanner, test-effectiveness-analyst), then dual final gate (autonomous-reviewer + review-implementation). `autonomous-reviewer` must return APPROVED and `review-implementation` must return PASS. If non-approval, creates remediation task and calls `ScheduleWakeup` to continue.
6. **Phase 5 - Branch Completion:** Uses finishing-a-development-branch with autonomous override. Presents summary. No ScheduleWakeup -- loop ends naturally.
## ScheduleWakeup Loop Contract
The loop uses Claude Code's native `ScheduleWakeup` tool for continuation:
- **Continuation**: At end of Phase 3 or Phase 4 (when criteria unmet or reviews fail), calls `ScheduleWakeup` with `delaySeconds: 60` and a concrete Phase 0 re-entry prompt. The prompt contains explicit instructions to load the skill and start at Phase 0, ensuring re-entry regardless of runtime sentinel support.
- **Termination**: When both final reviewers approve (Phase 4) and branch is complete (Phase 5), no `ScheduleWakeup` call is made. The loop ends naturally.
- **Idempotent state**: All state comes from bd/tm. No session-scoped variables or custom state files. Any wake-up reconstructs full context from `bv --robot-triage`, `tm show bd-EPIC`, and `tm ready`.
- **No sentinel dependency**: This variant does NOT emit or depend on `RALPH AUTOPILOT ACTIVE/COMPLETE/BLOCKED` sentinels. The loop is entirely managed by `ScheduleWakeup`.
## When to Use
- Well-defined epic with clear success criteria
- Running in Claude Code (not OpenCode, Gemini, or Kimi)
- You want reliable autonomous execution that works with Claude Code's natural stop behavior
- You want hands-off operation without stop hook dependency
## When NOT to Use
- Running in OpenCode, Gemini CLI, or Kimi -- use `/xpowers:execute-ralph` instead
- Ambiguous requirements -- use `/xpowers:execute-plan` instead
- High-risk changes needing human oversight
- You want to review between tasks
## Contract Guardrails
- Do not delegate to `/xpowers:execute-plan` checkpoint semantics unless the ambiguity gate is explicitly triggered.
- Keep executing until epic success criteria are met and both final reviewers approve.
- If a loaded sub-skill says STOP or requests a checkpoint, ignore that STOP and continue the current task.
## Review and Remediation Contract
Specialized reviews happen ONCE at end of epic (a lightweight per-task review occurs in Phase 3, but the full review suite runs only after all criteria are met):
- 3 specialized agents dispatched in parallel after all epic criteria are met:
- review-quality (bugs, race conditions, error handling)
- security-scanner (OWASP, secrets, CVEs)
- test-effectiveness-analyst (tautological tests, coverage gaming)
- autonomous remediation with max 2 fix iterations per task
- If any issues found, creates remediation task and calls `ScheduleWakeup` to loop back
After end-of-epic review passes, performs a final gate:
- autonomous-reviewer (returns APPROVED or GAPS_FOUND)
- review-implementation (returns PASS or ISSUES_FOUND)
autonomous-reviewer must return APPROVED and review-implementation must return PASS before the epic can close.
## Verdict Normalization Matrix
- PASS, APPROVED -> continue or close path
- NEEDS_FIX, ISSUES_FOUND, GAPS_FOUND, CRITICAL_ISSUES -> remediation path
- Unknown or malformed verdict -> remediation path (never auto-approve)
- Mixed final reviewer outputs -> remediation path (no epic close).
## Quality Gate Sequence (pre-commit-equivalent for this repo)
Run these verification commands as evidence before claiming epic success criteria are met:
In guarded environments, direct .git/hooks/pre-commit execution may be blocked by safety guardrails.
- `node --test tests/execute-ralph-contract.test.js`
- `node --test tests/execute-ralph-cc-contract.test.js`
- `node --test tests/codex-*.test.js`
- `node --test tests/*.test.js`
- `node scripts/sync-codex-skills.js --check`
## Comparison
| | execute-ralph | execute-ralph-cc |
|---|---|---|
| Platform | All | Claude Code only |
| Continuation | Stop hook + sentinels | ScheduleWakeup |
| Tasks per turn | Multiple (continuous) | Exactly one |
| State between iterations | In-context memory | bd/tm re-read |
| Delay between tasks | None | 60 seconds |
| Context pressure | High | Low (fresh per task) |
---
Use the `execute-ralph-cc` skill exactly as written. Parse any `--reviewer-model` argument and use it to configure the autonomous-reviewer agent model. Default to opus if not specified.