Code Explainer

This skill analyzes code (Python, JS, Java, etc.) and generates a professional, structured report to help the user understand its purpose and mechanics.

When to Use

Invoke this skill when the user:

• Asks "What does this code do?"
• Asks "Explain this file to me."
• Pastes a snippet and asks for an analysis.
• Wants to understand the purpose or design of a specific module.

Workflow

1. Context Loading: Read docs/AI_CONTEXT/ARCHITECTURE.md to understand the system context.

2. Read Code: Ensure you have the full content of the file or snippet the user is referring to.

3. Analyze:

   • Identify the main class/function/module.
   • Trace data flow and key logic.
   • Infer business intent from naming and comments.
   • Cross-Reference: Relate the code to Architecture decisions (ADR) if applicable.

4. Generate Report: Output a response strictly following the Report Format below.

Report Format

1. 核心功能 (Core Functionality)

[Summarize the main task of the code in 1-2 sentences. Example: "This script fetches data from API X, cleans it, and stores it in DB Y."]

2. 使用场景 (Use Cases)

[Describe when this code is used or what problem it solves in 1-2 sentences. Example: "Used during nightly batch jobs to synchronize local data with the remote server."]

3. 设计意图 (Design Intent)

[Explain the specific problem it addresses or the requirement it meets. Example: "Designed to handle network jitter by implementing exponential backoff retries."]

4. 关键逻辑解析 (Key Logic Breakdown)

[List the critical parts of the code and what they do.]

• [Function/Class Name]: [Description of what it implements]
• [Variable/Constant]: [Role in the logic]
• [Flow Control]: [Explanation of complex loops or conditions]

Example Output

1. 核心功能

这段代码的主要功能是解析本地的 JSON Registry 文件，并将其合并为一个统一的 llm_registry.json 构建产物。

2. 使用场景

通常在 CI/CD 流程中或发布新版本前使用，确保所有分散的厂商数据被正确打包分发给 SDK。

3. 设计意图

旨在解决单文件管理大量厂商数据带来的合并冲突问题，通过“分治”策略提升可维护性，同时保持对外交付物的单一性。

4. 关键逻辑解析

• load_providers(): 遍历 registry/providers/ 目录，加载所有独立的厂商配置。
• resolve_refs(): 递归处理 JSON 中的 $ref 字段，将引用的 Schema 文件内容内联到主文件中。
• merge_data(): 将处理后的数据组装成最终的字典结构，并写入磁盘。