eliteforge-tech-doc

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

EliteForge Tech Doc

EliteForge Tech Doc

目标

Objectives

  • 基于已有材料生成可评审技术设计文档,不凭空编造事实。
  • 覆盖当前执行目录中的全部文档与图:至少纳入来源清单并给出处置结论(已采纳/背景参考/待确认)。
  • 优先复用仓库中的既有术语、流程名称和图示风格。
  • Generate reviewable technical design documents based on existing materials, without fabricating facts out of thin air.
  • Cover all documents and diagrams in the current execution directory: at least include a source list and give disposal conclusions (Adopted/Background Reference/To Be Confirmed).
  • Prioritize reusing existing terminology, process names, and diagram styles in the repository.

执行顺序(每次都做)

Execution Sequence (Follow Every Time)

  1. 在当前执行目录运行
    scripts/list_doc_and_diagram_sources.sh .
    盘点资料。
  2. 建立“来源清单”并按优先级读取:
    • P0:
      README*
      *design*
      *architecture*
      *spec*
      docs/**
      中核心文档。
    • P1:接口、数据模型、部署、运维、故障处理文档。
    • P2:长尾说明文档与图像资源(至少做一行摘要,不遗漏)。
  3. 建立事实表(建议字段:
    结论
    证据文件
    置信度
    备注
    )。
  4. 先读 references/tech-doc-template.md,按模板起草目标文档。
  5. 优先复用已有图:
    • 存在
      .mmd/.mermaid/.puml/.plantuml/.drawio
      时,优先引用或改写为同等语义图块。
    • 仅存在图片图(
      .png/.jpg/.jpeg/.webp
      )时,只能依据周边文档和文件名归纳,不确定项标记为“待人工确认”。
  6. 输出文档到用户指定路径;用户未指定时默认写入当前目录
    ./tech-design-summary.md
  7. references/quality-checklist.md 自检后再回复结果。
  1. Run
    scripts/list_doc_and_diagram_sources.sh .
    in the current execution directory to inventory materials.
  2. Create a "Source List" and read materials by priority:
    • P0: Core documents in
      README*
      ,
      *design*
      ,
      *architecture*
      ,
      *spec*
      ,
      docs/**
      .
    • P1: Documents related to interfaces, data models, deployment, operation and maintenance, and fault handling.
    • P2: Long-tail explanatory documents and image resources (write at least one-line summary, no omissions).
  3. Create a Fact Table (suggested fields:
    Conclusion
    ,
    Evidence File
    ,
    Confidence Level
    ,
    Remarks
    ).
  4. First read references/tech-doc-template.md, then draft the target document according to the template.
  5. Prioritize reusing existing diagrams:
    • When
      .mmd/.mermaid/.puml/.plantuml/.drawio
      files exist, prioritize referencing or rewriting them into diagram blocks with equivalent semantics.
    • Only when image files (
      .png/.jpg/.jpeg/.webp
      ) exist, summarize based on surrounding documents and file names, and mark uncertain items as "To Be Confirmed Manually".
  6. Output the document to the path specified by the user; if no path is specified, write to the default path
    ./tech-design-summary.md
    in the current directory.
  7. Conduct a final check according to references/quality-checklist.md before replying with the result.

强制约束

Mandatory Constraints

  • 禁止臆造组件、接口、流程、状态、时序关系。
  • 每个关键结论至少绑定 1 个来源文件路径。
  • 发现冲突信息时,必须显式列出冲突点与取舍依据。
  • 不得忽略扫描结果中的任何文档/图文件;无法消费的文件必须写明原因。
  • 输出文档必须包含“来源清单”和“待确认项”。
  • Forbid inventing components, interfaces, processes, states, or sequence relationships.
  • Each key conclusion must be bound to at least one source file path.
  • When conflicting information is found, explicitly list the conflict points and basis for trade-offs.
  • Do not ignore any documents/diagram files in the scan results; must state the reason for files that cannot be processed.
  • The output document must include "Source List" and "Items To Be Confirmed".

图与流程提炼规则

Diagram and Process Extraction Rules

  1. 对流程类内容,优先输出“场景化结构”:触发条件、输入、步骤、输出、异常分支。
  2. 对调用链内容,优先使用
    mermaid
    sequenceDiagram
    表达关键参与者与调用方向。
  3. 对状态流转内容,优先使用
    mermaid
    stateDiagram-v2
    表达状态与迁移条件。
  4. 对伪代码,保留领域动作名,不替换为空泛描述。
  5. 对无法确认的图含义,保留原文件路径并标注“待人工确认”,不要猜测细节。
  1. For process-related content, prioritize outputting "scenario-based structure": trigger conditions, inputs, steps, outputs, and exception branches.
  2. For call chain content, prioritize using
    mermaid
    sequenceDiagram
    to express key participants and call directions.
  3. For state transition content, prioritize using
    mermaid
    stateDiagram-v2
    to express states and transition conditions.
  4. For pseudo-code, retain domain action names and do not replace them with vague descriptions.
  5. For diagrams with unclear meanings, retain the original file path and mark it as "To Be Confirmed Manually"; do not guess details.

产出文档最低结构

Minimum Structure of Output Document

按模板输出,至少包含以下章节:
  1. 文档来源清单
  2. 背景与目标
  3. 术语与边界
  4. 架构概览
  5. 核心场景流程(可含伪代码)
  6. 时序图
  7. 状态图
  8. 数据与接口要点
  9. 风险与待确认项
优先采用“交互概览 -> 场景化流程(伪代码)-> 时序图 -> 状态图”的组织方式,便于从业务动作追溯到系统实现。
Output according to the template, including at least the following sections:
  1. Document Source List
  2. Background and Objectives
  3. Terminology and Boundaries
  4. Architecture Overview
  5. Core Scenario Processes (may include pseudo-code)
  6. Sequence Diagram
  7. State Diagram
  8. Key Points of Data and Interfaces
  9. Risks and Items To Be Confirmed
Prioritize the organization method of "Interactive Overview -> Scenario-based Process (Pseudo-code) -> Sequence Diagram -> State Diagram" to facilitate tracing from business actions to system implementation.

输出风格

Output Style

  • 默认中文输出,关键名词沿用原文(如模块名、接口名、类名)。
  • 段落短句化,避免泛化空话。
  • 图和伪代码只保留对评审决策有价值的内容,避免冗长重复。
  • Default output in Chinese, retain original terms for key nouns (such as module names, interface names, class names).
  • Use short paragraphs, avoid vague and empty statements.
  • Only retain content valuable for review decisions in diagrams and pseudo-code; avoid lengthy repetitions.

参考资料读取策略

Reference File Reading Strategy

  • 首次执行先读 references/tech-doc-template.md
  • 出稿前必读 references/quality-checklist.md 做最终检查。
  • Read references/tech-doc-template.md first during the first execution.
  • Read references/quality-checklist.md for final check before delivering the draft.

快速命令

Quick Commands

bash
undefined
bash
undefined

1) 盘点当前目录文档和图

1) Inventory documents and diagrams in the current directory

skills/eliteforge-tech-doc/scripts/list_doc_and_diagram_sources.sh .
skills/eliteforge-tech-doc/scripts/list_doc_and_diagram_sources.sh .

2) 生成技术设计文档(文件名可改)

2) Generate technical design document (file name can be modified)

输出示例:./tech-design-summary.md

Output example: ./tech-design-summary.md

undefined
undefined

示例触发语句

Example Trigger Phrases

  • “请扫描当前目录所有文档和图,产出技术设计文档。”
  • “根据仓库里的 markdown 和时序图,整理一份方案设计稿。”
  • “把现有设计资料汇总成可评审的 tech design,并标出缺失信息。”
  • "Please scan all documents and diagrams in the current directory and produce a technical design document."
  • "Based on the Markdown files and sequence diagrams in the repository, organize a solution design draft."
  • "Summarize the existing design materials into a reviewable tech design and mark missing information."

参考文件

Reference Files

  • 技术设计文档模板
  • 质量自检清单
  • Technical Design Document Template
  • Quality Self-Check Checklist