菜单
End-to-end ship pipeline — commit, push, PR, 10-step review, fix, CI, merge. Use on "ship it", "done", "merge it", "push it", "deliver this".
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
| name | shipping-workflow |
| description | End-to-end ship pipeline — commit, push, PR, 10-step review, fix, CI, merge. Use on "ship it", "done", "merge it", "push it", "deliver this". |
| license | MIT |
| metadata | {"author":"summerela","version":"1.2.0"} |
Apply token-economy whenever this skill would otherwise trigger broad
exploration, repeated file reads, multi-file scans, or heavy reference
loading.
End-to-end ship pipeline: commit, push, PR, review, CI, merge, cleanup.
This file is a router; detailed review and branch-handling doctrine stays
in phases and reference.md.
claude-flow before shippingWhen /next (or any caller) ships a session-handoff doc, skip review/CI and auto-merge to main. Future sessions need handoff docs discoverable on main, not stranded in open PRs.
Conditions for auto-merge (ALL must hold):
gh pr view <N> --json files --jq '.files | length' returns 1)docs/plans/*handoff*.md (case-insensitive handoff substring in the basename)changeType: "ADDED" OR (changeType: "MODIFIED" AND the file path matches docs/plans/*handoff*.md AND deletions <= 5). Rationale: refreshing the bottom continuation prompt of an existing handoff doc to reflect current shipped/in-flight state commonly produces small deletions (the prior prompt is rewritten in place). courierflow_beta PR #40 shipped at +91 / -3 (the 3 deletions were the stale bottom prompt) — well within spirit. The <=5 cap keeps substantive rewrites on the normal review path. Substantive prompt rewrites that delete more than 5 lines (e.g. courierflow_beta PR #50 at +37 / -7 — the deletions were a continuation-prompt rewrite reflecting an intervening merge) intentionally fall through to normal review; the cap is not raised for them, because rewriting the next-session entry point deserves a human read.docs/...)mainWhen all hold: after Phase 2 opens the PR, immediately run:
env -u GH_TOKEN gh pr merge <N> --squash --delete-branch
Skip Phase 3 (review + CI) entirely. Phase 4 verifies the merge state and hands off to /cleanup as usual.
When any condition fails: fall through to the normal Phase 3 review flow — the PR is treated as ordinary doc work, not a handoff.
Why narrow: the conditions exist to catch the exact handoff-doc shape (/next writes single-file additive docs under docs/plans/*handoff*.md). A multi-file PR, a doc that also touches code, a non-docs/ branch, or a deletion-bearing diff all fall back to the normal review path. The fast path never auto-merges substantive changes.
Why auto-merge is safe for this case: handoff docs are write-once continuation prompts. No executable code, no schema, no security surface. The review gate exists to catch issues in code/config changes — for doc-only continuation prompts it just delays the next session from finding the doc. Validated 2026-05-22 on courierflow_beta PR #18 (Task 4 handoff doc) which sat open until manually merged into commit 598fd11 — exactly the failure mode this fast path prevents.
Auto-mode classifier blocks the fast-path merge even when all conditions match. In auto-mode sessions, the harness classifier flags gh pr merge <N> --squash --delete-branch as "self-squash-merging to main without explicit user authorization" and pauses for approval — even when the fast-path's five conditions all hold. The skill's fast-path logic doesn't translate to the classifier (there's no per-PR signal saying "this is a handoff doc, fast-path eligible"). When the merge is blocked: surface a one-line summary to the user ("PR #N matches handoff-doc fast path — approve merge?") rather than falling through to Phase 3 review. Hit 2026-05-22 on courierflow_beta PR #25. Long-term fix is a classifier-side signal — out of scope for this skill.
Update 2026-05-28 — validated on courierflow_beta PR #109 (handoff doc docs/plans/2026-05-28-correctness-rubric-phase5d1-handoff.md on branch docs/correctness-rubric-phase5b-handoff): fast-path merge executed in auto-mode WITHOUT classifier interception. All five fast-path conditions held (single file, docs/plans/*handoff*.md, ADDED, doc-prefixed branch, base=main). Either the classifier was updated to recognize the handoff-doc shape, or the trigger surface is narrower than originally documented. Refined guidance: attempt the fast-path merge directly when all five conditions hold; if the classifier blocks, fall through to the one-line user surface as before. Don't pre-emptively skip the merge attempt on the assumption that it will be blocked.
Some repos (notably docs-style ones — claude-skills is the canonical example) have no CodeRabbit, no .github/workflows/, and no human reviewers attached to PRs. A PR there is effectively ready-to-merge the moment it opens; waiting for review feedback that will never come is dead time and confuses handoff docs that assume CR is a gate.
Detection (run before entering Phase 3):
env -u GH_TOKEN gh pr view <recent-merged-PR-#1> --json reviews,statusCheckRollup
env -u GH_TOKEN gh pr view <recent-merged-PR-#2> --json reviews,statusCheckRollup
Pick the two most recently merged PRs in the same repo. If reviews is [] AND statusCheckRollup is [] on BOTH → no review wiring exists. Confirm by checking the tree: ls .github/workflows/ 2>/dev/null returns nothing and there's no .coderabbit.yaml / .coderabbit.yml.
When detected: skip Phase 3's wait-for-review step. Phase 3's self-review checklist (the "10-step review") still runs as a self-audit, but stops blocking on external signal. Proceed to Phase 4 merge as soon as self-review passes.
Explicit no-CI approval still enters cleanup. If Summer explicitly approves a no-CI merge for a repo with no reported checks, merge with the required approval marker, verify the merge, then continue to Phase 4 cleanup. The no-CI approval replaces only the external CI/review gate; it does not skip session-learnings, repo sync, or worktree teardown.
Distinct from the handoff-doc fast path above: that one fires on PR shape (single-file, additive, docs/plans/*handoff*.md). This one fires on REPO shape (no review automation). They compose — a handoff doc in a no-CR repo hits both.
Why this matters: validated 2026-05-27 on claude-skills PR #118 — handoff doc treated "address CodeRabbit feedback" as a gating Phase 3 step. CodeRabbit had never been wired up in this repo; PRs #117 and #114 each had reviews: [] and statusCheckRollup: []. Without this detection step, the resumer waits indefinitely for review that won't fire.
Composes with next skill's §5a "Re-verify on resume" — that section makes the CR/CI check part of the resumer's preflight, so the no-review-wired-up state is known before Phase 3 starts.
phases/.reference.md only when running or adapting the review stage.phases/phase-1-preflight-and-commit.mdphases/phase-2-push-and-pr.mdphases/phase-3-review-and-ci.mdphases/phase-4-merge-and-cleanup.md/cleanup owns final session-learnings, sync, and teardown.Produce only what the current shipping phase needs:
Conservative cross-repo worktree + branch pruning audit. Classifies every branch/worktree as safe-to-delete (proven dead via merged PR or zero unique commits) or keep-with-reason, then deletes what's proven dead. Use for 'clean up stale branches', 'prune dead worktrees', 'repo hygiene sweep', or any multi-repo branch-pruning ask.
Best practices for AI-authored writing — consolidated rules for when Claude (or any LLM) drafts prose a human will put their name on: emails, posts, newsletters, docs, scripts, bios, marketing copy. Use whenever AI-drafted output must not read as generic AI sludge; triggers on '/ai-writing', 'de-slop this draft', 'make this not sound like AI'. Pulls the AI-relevant rules from the writing and communication skills into one pre-ship gate. NOT a voice profile (use writing-voice for Summer's voice, sme-voice for someone else's) and NOT a conflict/timing layer (use communication-safeguards for heated messages).
Document a dead end so future sessions avoid re-exploring. Writes an abandon record to .claude/abandoned/<date>-<topic>.md capturing what was tried, why it failed, and what was learned. Invoked as `/session-handoff --abandon`, or when the user says "abandon this", "this didn't work", "scrap this approach", or "dead end". Also used by claude-flow Phase 1 (Explore/Archive path) and the research skill. Normal state-export handoffs are written by the `next` skill to docs/plans/<date>-session-handoff.md — not by this skill.
Captures another person's writing voice from samples and applies it when ghostwriting or editing in their voice — primarily for SME (subject-matter expert) scripts at DeepLearning.AI. Phase A builds a saved voice profile from samples; Phase B applies it when editing or extending the SME's scripts. Profiles live in ~/claude_code/agent-vault/sme-voices/ to sync across machines. Trigger on 'capture [name]'s voice', 'edit this in [name]'s voice', 'ghostwrite as [name]', or '/sme-voice'. NOT for writing as Summer herself (use writing-voice), ToneGuard analysis, or summarizing/analyzing documents.
Typography rules — quote marks, dashes, spacing, hierarchy, HTML entities. Auto-apply when generating HTML/CSS/React/Jinja with visible text; audit on "fix the typography", "typography audit", "make this look professional".
Loads Summer Rae's personal voice profile for any writing she'll put her name on — fiction, personal essays, blog posts, newsletters, social posts, course marketing hooks, L0 scripts, business emails and documents. Applies her creative voice (sensory, deadpan dark humor, show-don't-tell) and her business-writing preferences (minimal, direct, lead with the ask, no hollow openers). Default whenever Summer asks you to write or draft anything ('write this for me', 'in my voice'). NOT for ghostwriting or preserving an SME's voice (use sme-voice), ToneGuard analysis, or summarizing/analyzing documents.