T
thetonymaster/composition-software-engineering
Claude Code plugin marketplace: design-principle skills (composition, dependency injection, flat code,…
无 SECURITY.md
在 GitHub 查看 ↗链接已于 2026-11-12 核查 · HTTP 200 · 打开的是厂商自己的官网
- ★ 数取值
- 0 ★ · thetonymaster/composition-software-engineering · 取自 GitHub 搜索 API · (2026-11-13 20:00:50 核对)
- 来源明细
- 从该仓库的 GitHub 页面直接读取 · 条目类型:开源项目
- 收录与活跃
- 收录于 2026-10-23 · 最近见到 2026-11-14 · 共见到 26 次 · 被 2 个来源列出
- 评分构成
- 综合分 203 · 2 个来源 · repo
常见问题(答案都来自我们的记录)
它适合谁 / 用来做什么?
我们把它归在这些标签下: 开源项目。标签来自条目自身,不是我们强加的分类。
谁维护它、还在更新吗?
仓库
thetonymaster/composition-software-engineering;我们收录于 2026-10-23;最近见到 2026-11-14。这里没有"官方认证"或"质量评级"——我们只记录来源与时间。# composition-software-engineering
以下摘录自该仓库的 README(片段) · 在 GitHub 读完整 README →
A Claude Code plugin marketplace with one plugin, principles: five skills
that steer how Claude designs, refactors and reviews code, and one agent that
audits code against them.
| Skill | Use it when |
| composition | reaching for an abstraction, a base class, a mixin, embedding or use; deciding whether an abstraction earns its keep |
| dependency-injection | code hardcodes a client, global, disk, network or clock; tests need monkeypatching; one function branches per backend |
| flat-code | a function nests past depth 3, chains if/else, or has an else after return |
| measure-first | someone says "make it faster", "optimize" or "is X faster than Y", before trading readability for speed |
| principles-audit | auditing a diff or module against the four principles above, or briefing a subagent to work under them |
| Agent | Use it when |
| principles-auditor | you want a principles-audit report from a fresh context that did not write the code; it reads and runs checks, and does not edit |
Every skill has references for Go, Python, Elixir, TypeScript, Java, Rust
and C# (references/ .md), loaded only when the code under review
is in that language. flat-code ships scripts/nestingdepth.py, which
measures nesting in all seven (Python exactly from the ast, the rest by an
indentation heuristic) and exits 2 on any other file type. Against exact
parsers the heuristic found 87/90 over-nested TypeScript functions and 59/60
Java methods; Rust and C# were spot-checked by hand.
## Install
From GitHub (needs access to this repository):
/plugin marketplace add thetonymaster/composition-software-engineering
/plugin install principles@composition-software-engineering
From a local clone:
/plugin marketplace add /path/to/composition-software-engineering
/plugin install principles@composition-software-engineering
Skills are namespaced once installed: principles:composition,
principles:flat-code, and so on. They trigger from their descriptions; you
can also invoke one by name, e.g. /principles:principles-audit. The agent
is principles:principles-auditor; ask Claude to "run the principles-auditor
agent on this branch", or pick it from /agents.
## How much of the reference material was executed
| Language | Snippets | Tool names, flags, rule IDs |
| Go | key claims probed (embedding, nil interface, b.Loop benchmark); not every snippet run | audit greps run on a real repo |
| Python | not executed | reviewed; profiling.sampling CLI checked |
| Elixir | not executed (Benchee example not run) | reviewed |
that steer how Claude designs, refactors and reviews code, and one agent that
audits code against them.
| Skill | Use it when |
| composition | reaching for an abstraction, a base class, a mixin, embedding or use; deciding whether an abstraction earns its keep |
| dependency-injection | code hardcodes a client, global, disk, network or clock; tests need monkeypatching; one function branches per backend |
| flat-code | a function nests past depth 3, chains if/else, or has an else after return |
| measure-first | someone says "make it faster", "optimize" or "is X faster than Y", before trading readability for speed |
| principles-audit | auditing a diff or module against the four principles above, or briefing a subagent to work under them |
| Agent | Use it when |
| principles-auditor | you want a principles-audit report from a fresh context that did not write the code; it reads and runs checks, and does not edit |
Every skill has references for Go, Python, Elixir, TypeScript, Java, Rust
and C# (references/ .md), loaded only when the code under review
is in that language. flat-code ships scripts/nestingdepth.py, which
measures nesting in all seven (Python exactly from the ast, the rest by an
indentation heuristic) and exits 2 on any other file type. Against exact
parsers the heuristic found 87/90 over-nested TypeScript functions and 59/60
Java methods; Rust and C# were spot-checked by hand.
## Install
From GitHub (needs access to this repository):
/plugin marketplace add thetonymaster/composition-software-engineering
/plugin install principles@composition-software-engineering
From a local clone:
/plugin marketplace add /path/to/composition-software-engineering
/plugin install principles@composition-software-engineering
Skills are namespaced once installed: principles:composition,
principles:flat-code, and so on. They trigger from their descriptions; you
can also invoke one by name, e.g. /principles:principles-audit. The agent
is principles:principles-auditor; ask Claude to "run the principles-auditor
agent on this branch", or pick it from /agents.
## How much of the reference material was executed
| Language | Snippets | Tool names, flags, rule IDs |
| Go | key claims probed (embedding, nil interface, b.Loop benchmark); not every snippet run | audit greps run on a real repo |
| Python | not executed | reviewed; profiling.sampling CLI checked |
| Elixir | not executed (Benchee example not run) | reviewed |
收录徽章(可选)
想把收录标记放进 README 就复制这行,可选,不影响收录。
[](https://anclum.com/zh/s/thetonymaster-composition-software-engineering/)
- 来源
- github · https://github.com/thetonymaster/composition-software-engineering
- 按领域浏览
- 开源项目
- 收录来源
- 2 个独立来源
- 收录于
- 今天
- 安全策略
- 我们没在该仓库根目录找到
SECURITY.md(这不代表不安全)
同类推荐
所属领域与标签
每个标签都能点进去看同类的全部条目。数字是当前库里的真实条数。