Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/vinvcn/addyosmani-agent-skills-zhWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/vinvcn/addyosmani-agent-skills-zh/ship)<a href="https://agentmods.dev/commands/vinvcn/addyosmani-agent-skills-zh/ship"><img src="https://agentmods.dev/badge/commands/vinvcn/addyosmani-agent-skills-zh/ship/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/vinvcn/addyosmani-agent-skills-zh/ship"><img src="https://agentmods.dev/badge/commands/vinvcn/addyosmani-agent-skills-zh/ship.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00019 | $0.01069 |
| Opus 5 | $0.00010 | $0.00535 |
| Sonnet 5 | $0.00004 | $0.00214 |
| Haiku 4.5 | $0.00002 | $0.00107 |
Grade A, and why
ship scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 73 lines — stays where its author put it; the contents beside it link to each section on GitHub.
调用 agent-skills:shipping-and-launch skill。
/ship 是一个 fan-out orchestrator。它会针对当前变更并行运行三个 specialist personas，然后将它们的报告合并为一个带 rollback plan 的 go/no-go decision。Personas 独立运行，没有共享状态，也没有顺序依赖；这正是此处并行执行安全且有用的原因。
Phase A — Parallel fan-out
使用 Agent tool 并发 spawn 三个 subagents。在同一个 assistant turn 中发出全部三个 Agent tool calls，让它们并行执行；顺序调用会违背此 command 的目的。
在 Claude Code 中，每次调用都传入与 persona 的 name 字段匹配的 subagent_type：
code-reviewer— 对 staged changes 或 recent commits 执行五轴 review（correctness、readability、architecture、security、performance）。输出标准 review template。security-auditor— 执行 vulnerability 和 threat-model pass。检查 OWASP Top 10、secrets handling、auth/authz、dependency CVEs。输出标准 audit report。test-engineer— 分析该变更的 test coverage。识别 happy path、edge cases、error paths 和 concurrency scenarios 中的 gaps。输出标准 coverage analysis。
在没有 Agent tool 的其他 harnesses 中，顺序调用每个 persona 的 system prompt，并将它们的输出视作并行返回；merge phase 仍然有效。
约束（来自 Claude Code 的 subagent model）：
- Subagents 不能 spawn 其他 subagents；不要让一个 persona delegate 给另一个。
- 每个 subagent 都有自己的 context window，并且只向这个 main session 返回自己的 report。
- 如果你需要 teammates 互相对话，而不只是回传报告，请使用 Claude Code Agent Teams，并将这些 personas 引用为 teammate types（见
references/orchestration-patterns.md）。
Persona resolution。 如果你在 .claude/agents/ 或 ~/.claude/agents/ 中定义了自己的 code-reviewer、security-auditor 或 test-engineer，它们会优先于此 plugin 的版本；/ship 会自动采用你的自定义内容。这是有意设计的：plugin subagents 位于 Claude Code scope priority table 的底部，因此 user-level definitions 会按设计优先。
Phase B — Merge in main context
三个 reports 都返回后，main agent（不是 sub-persona）进行综合：
- Code Quality — 汇总
code-reviewer的 Critical/Important findings，以及任何 failing tests、lint 或 build output。合并 reviewers 之间的重复项。 - Security — 将任何 Critical/High
security-auditorfindings 提升为 launch blockers。与code-reviewer的 security axis 交叉核对。 - Performance — 从
code-reviewer的 performance axis 提取；如适用，交叉检查 Core Web Vitals。 - Accessibility — 验证 keyboard nav、screen reader support、contrast（三个 personas 不覆盖这一项；在这里直接处理，或调用 accessibility checklist）。
- Infrastructure — Env vars、migrations、monitoring、feature flags。直接验证。
- Documentation — README、ADRs、changelog。直接验证。
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 73 lines · 19 tokens per session scan A a0d7643b883e
ship is a command published in the GitHub repository vinvcn/addyosmani-agent-skills-zh (31 stars, last pushed 4mo ago), licensed MIT. It adds 19 tokens to every session and 1,069 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
ship
Run the pre-launch checklist via parallel fan-out to specialist personas, then synthesize a go/no-go decision.
ship
Run the pre-launch checklist via parallel fan-out to specialist personas, then synthesize a go/no-go decision.
build
Implement tasks incrementally — build, test, verify, commit. Add "auto" to run the whole plan in one approved pass.
build
Implement tasks incrementally — build, test, verify, commit. Add "auto" to run the whole plan in one approved pass.
release
Drive the release lifecycle from one command — verify CI and ship, or cut a versioned tag with patch notes — instead of remembering which release skill fits which step. Trunk-based: releases are tags cut from the trunk (the repo's default branch); staging and production are deployment environments driven by CI/CD and…
plan
Break work into small verifiable tasks with acceptance criteria and dependency ordering.