使用方法
- 复制安装命令,在自己的智能体环境安装前先检查源码包。
- 从原始文档核对所需工具、凭据和支持的宿主。
- 说明输入材料与期望交付物,并对照来源检查输出。
npx skills add https://github.com/afaraha8403/balakit --skill subsystem-walkthrough开始前需了解
- 安装不包含模型订阅、API 额度或访问文件的权限。
- 仓库星数和安装量反映活跃情况,不等于独立质量或安全测试。
按源仓库内容呈现,保留标题、案例、代码、表格、链接以及原文引用的演示图片。
Subsystem walkthrough
Leading words: mental model, one explainer, parallel slices, inherit host model.
Answer "how does X work?" at the level of a senior engineer onboarding onto the subsystem. Not annotated source.
Copy these steps into the todo list verbatim. Skip with skip: <reason>.
- Assess complexity. Simple → one explainer. Complex → 2–4 read-only explorers, then one synthesizer. Write
Complexity: simpleorComplexity: complex (N slices)as the first line of the output. - Explore or explain (see below). Inherit the host model. Do not hardcode a slug.
- Present the explainer output with light edits only. Keep the Complexity line.
When in doubt, take the simple path.
Simple
One read-only subagent explores and explains in a single pass. Fill references/explainer-prompt.md without explorer findings.
Complex
Decompose into 2–4 distinct slices. Spawn explorers in one message, each with references/explorer-prompt.md. Then one synthesizer with the explainer template and every explorer's findings filled in. If the host cannot spawn subagents, run the slices sequentially yourself.
Output
First line: Complexity: simple or Complexity: complex (N slices).
Then: Overview / Key Concepts / How It Works / Where Things Live / Placement / Gotchas. Drop a section that does not apply. Placement is where new code for this subsystem should live (package, file, layer) — one short paragraph, not a second architecture essay.
Reply contract
The Complexity line, then the explanation. Do not rewrite it into a different structure after the explainer returns.
