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.
npx skills add gongyijie85/mattpocock-skills-dsh-zh --skill improve-codebase-architecture-zhgit clone --depth 1 https://github.com/gongyijie85/mattpocock-skills-dsh-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/skills/gongyijie85/mattpocock-skills-dsh-zh/improve-codebase-architecture-zh)<a href="https://agentmods.dev/skills/gongyijie85/mattpocock-skills-dsh-zh/improve-codebase-architecture-zh"><img src="https://agentmods.dev/badge/skills/gongyijie85/mattpocock-skills-dsh-zh/improve-codebase-architecture-zh/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/skills/gongyijie85/mattpocock-skills-dsh-zh/improve-codebase-architecture-zh"><img src="https://agentmods.dev/badge/skills/gongyijie85/mattpocock-skills-dsh-zh/improve-codebase-architecture-zh.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 | $0.00052 | $0.01692 |
| Opus 5 | $0.00026 | $0.00846 |
| Sonnet 5 | $0.00010 | $0.00338 |
| Haiku 4.5 | $0.00005 | $0.00169 |
Grade A, and why
improve-codebase-architecture-zh 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 yesterday.
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 — 72 lines — stays where its author put it; the contents beside it link to each section on GitHub.
改进代码库架构
找出架构摩擦点,并提出 deepening opportunities(加深机会)——把浅层模块变成深层模块的 refactor。目标是可测试性与 AI 可导航性(AI-navigability)。
该命令以项目的领域模型为_依据_,并建立在一套共享的设计词汇之上:
- 调用 skill tool 的 "codebase-design" 获取架构词汇(module、interface、depth、seam、adapter、leverage、locality)及其原则(deletion test(删除测试)、"the interface is the test surface"(接口即测试表面)、"one adapter = hypothetical seam, two = real"(一个 adapter = 假想的 seam,两个 = 真实的))。每条建议中都精确使用这些术语——不要滑向 "component"、"service"、"API" 或 "boundary"。
CONTEXT.md中的领域语言为好的 seams 命名;docs/adr/中的 ADRs 记录了该命令不应重新争论(re-litigate)的决策。
流程
1. 探索
先界定范围再扫描——YAGNI。 加深一个模块的价值在于让未来对它的修改更容易,因此要把额外权重放在代码库中最近有变动的部分。在动手看之前先决定看哪里:
- 如果用户指明了方向——某个模块、子系统或痛点——直接采纳,跳过下面的推断。
- 否则,往回翻阅一段较长的提交历史(
git log --oneline),找出代码库的 hot spots(热点)——反复出现的文件和区域——让这些路径先吸引你的注意力。如果变更分散、没有明确热点,就扩大搜索范围。
先阅读项目的领域词汇表(CONTEXT.md)以及你要触及区域中的任何 ADRs。
然后派生一个 sub-agent 去遍历代码库。不要遵循僵化的启发式规则——有机地探索,并记录你在哪里感受到摩擦:
- 在哪里,理解一个概念需要在许多小模块之间来回跳转?
- 哪些模块是 shallow(浅层) 的——interface 几乎与实现一样复杂?
- 哪些地方仅仅为了可测试性就抽取了纯函数,而真正的 bug 却藏在调用方式里(没有 locality(局部性))?
- 哪些紧耦合的模块在 seams 之间泄漏?
- 代码库的哪些部分未被测试,或难以通过它们当前的 interface 测试?
对你怀疑是 shallow 的任何东西应用 deletion test(删除测试):删除它会集中复杂性,还是只是转移它?「会集中」正是你想要的信号。
2. 以 HTML 报告形式呈现候选
把自包含的 HTML 文件写入操作系统的临时目录,这样仓库里不会留下任何东西。从 $TMPDIR 解析临时目录,回退到 /tmp(Windows 上是 %TEMP%),写入 <tmpdir>/architecture-review-<timestamp>.html,这样每次运行都会得到新文件。为用户打开它——Linux 上用 xdg-open <path>,macOS 上用 open <path>,Windows 上用 start <path>——并告诉他们绝对路径。
报告用 通过 CDN 引入的 Tailwind 做布局与样式,并在图/流程/时序能可靠传达结构的地方用 通过 CDN 引入的 Mermaid 画图。把 Mermaid 与手工制作的 CSS/SVG 视觉元素混用——关系呈图状时(调用图、依赖、时序)用 Mermaid;想要更有编辑感的东西时(体量图、剖面图、折叠动画)用手工构建的 div/SVG。每个候选都配一个 before/after(改造前/后)可视化。要注重视觉化。
为每个候选渲染一张卡片,包含:
- Files(涉及文件) — 涉及哪些文件/模块
- Problem(问题) — 当前架构为何造成摩擦
- Solution(方案) — 用通俗语言描述会发生什么变化
- Benefits(收益) — 从 locality 与 leverage 的角度解释,以及测试将如何改善
- Before / After 示意图 — 并排、手工绘制,展示 shallow 与加深的效果
- Recommendation strength(推荐强度) —
Strong、Worth exploring、Speculative之一,渲染为徽章
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.
- yesterday First seen · 72 lines · 52 tokens per session scan A 08b663468015
improve-codebase-architecture-zh is a skill published in the GitHub repository gongyijie85/mattpocock-skills-dsh-zh (5 stars, last pushed 2d ago), licensed MIT. It adds 52 tokens to every session and 1,692 once invoked, about $0.0003 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 2026-09-11.
Other skills, from other repositories
code-review-cn
A Chinese-language guide for reviewing code changes, pull requests, and diffs. It prioritizes correctness, resource cleanup, security, matching documentation, and tests before style.
pr-description
A writing guide for pull request descriptions. A pull request is a proposed set of code changes for review and merging.
refactor-safely
A safe code-refactoring guide for changing code structure while keeping its external behavior the same. Refactoring means reorganizing code without changing what it does.
delivery-review
Adversarial self-review before delivery. Use once the implementation reaches green and before you declare the work done — assume the delivery fails its own spec, hunt for the strongest supportable objections, answer them, and re-review after fixes.
code2skill-review-source
A read-only review skill for checking whether a Code2Skill-generated result matches the source code it was authorized to use. It examines request handling, tool handoffs, transformations, authentication, and attachments.
dsh-pr-review
A checklist and feedback format for reviewing a pull request, which is a proposed set of code changes before they are added to a project.