Token导航 LogoToken导航

Agent Skill · 研究检索

feature-docs(功能文档)

用于辅助文档、README、Markdown、说明文和内容稿件的整理与改写。它适合让 Agent 提炼结构、补齐章节、统一术语、检查链接或把零散材料整理成可读文档。使用时应保留项目已有事实、命令和路径,不要把未确认的信息写成确定结论;涉及对外文案时,还需要控制语气,避免过度营销或夸大能力。

feature-docs 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证为 MIT,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 45,481。

总安装 847周安装 36GitHub Stars 6下载 297核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:feature-docs(功能文档)
来源仓库:https://github.com/duc01226/easyplatform
仓库路径:skills/feature-docs
安装命令:
npx skills add https://github.com/duc01226/easyplatform --skill feature-docs
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/duc01226/easyplatform --skill feature-docs

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

用于生成全面的业务功能文档,要求每个测试用例都有代码证据支撑。自动检测模块变更或指定模块,收集现有文档和代码信息。按固定 17 章节结构编写文档,严格遵循证据驱动原则。适合需要高质量技术文档输出的企业级应用场景。feature-docs 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

通过 GitHub 安装,支持多种 AI 编程助手环境集成。

来源仓库内路径:skills/feature-docs

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/duc01226/easyplatform --skill feature-docs;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

feature-docs 怎么安装?

在终端执行 npx skills add https://github.com/duc01226/easyplatform --skill feature-docs。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

feature-docs 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

feature-docs 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

feature-docs 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

feature-docs 用的人多吗?

公开数据是累计安装 847 次,近一周安装 36 次,下载 297 次,GitHub 6 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

feature-docs 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 3 个目录同时收录,说明这个 Skill 流传比较广。
  • 许可证明确为 MIT,商用和分发边界好判断。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
feature-docs

功能文档 · 本页

8476需要联网MIT通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

[IMPORTANT] Use TaskCreate to break ALL work into small tasks BEFORE starting — including tasks for each file read. This prevents context loss from long files. For simple tasks, AI MUST ATTENTION ask user whether to skip.

Quick Summary

Goal: Generate comprehensive 17-section business feature documentation with mandatory code evidence for all test cases.

Workflow:

  1. Detect & Gather — Auto-detect modules from git changes OR user-specified module, read existing docs
  2. Investigate Code — Grep/glob codebase to gather evidence (file:line format) for every test case
  3. Write Documentation — Follow exact 17-section structure, place in docs/business-features/{Module}/
  4. Verification — 3-pass system: evidence audit, domain model verification, cross-reference audit

Key Rules:

  • EVERY test case MUST ATTENTION have verifiable code evidence (FilePath:LineNumber), no exceptions
  • Output must have exactly 17 sections matching the master template
  • Always update CHANGELOG.md and Version History (Section 17) when modifying docs
  • When writing Section 15 test cases: include an IntegrationTest field pointing to the test file and method name. Format: IntegrationTest: Orders/OrderCommandIntegrationTests.cs::{MethodName}. If no integration test exists yet, set Status: Untested.
  • Verify every TC-{FEATURE}-{NNN} in Section 15 has a corresponding [Trait("TestSpec", "TC-{FEATURE}-{NNN}")] in the integration test codebase. If missing, flag as Status: Untested.
  • If third verification pass finds >5 issues, HALT and re-run verification

Be skeptical. Apply critical thinking, sequential thinking. Every claim needs traced proof, confidence percentages (Idea should be more than 80%).

  • docs/project-reference/domain-entities-reference.md — Domain entity catalog, relationships, cross-service sync (read when task involves business entities/models) (content auto-injected by hook — check for [Injected:...] header before reading)

Project Pattern Discovery

Before implementation, search your codebase for project-specific patterns:

  • Search for: business-features, detailed-features, feature-docs-template
  • Look for: existing feature doc folders, 17-section templates
MANDATORY IMPORTANT MUST ATTENTION Read the feature-docs-reference.md companion doc for project-specific patterns and code examples. If file not found, continue with search-based discovery above.

Feature Documentation Generation & Verification

Generate comprehensive feature documentation following project conventions and folder structure.

GOLD STANDARD References:

Search your codebase for existing feature docs to use as reference:

find docs/business-features -name "README.*.md" -type f | head -5

Template File: docs/templates/detailed-feature-docs-template.md


[CRITICAL] MANDATORY CODE EVIDENCE RULE

EVERY test case MUST ATTENTION have verifiable code evidence. This is non-negotiable.

Evidence Format

**Evidence**: `{RelativeFilePath}:{LineNumber}` or `{RelativeFilePath}:{StartLine}-{EndLine}`

Valid vs Invalid Evidence

ValidInvalid
ErrorMessage.cs:83{FilePath}:{LineRange} (template)
Handler.cs:42-52SomeFile.cs (no line)
interviews.service.ts:115-118"Based on CQRS pattern" (vague)

Output Structure

All documentation MUST ATTENTION be placed in the correct folder structure:

docs/
├── BUSINESS-FEATURES.md              # Master index (UPDATE if new module)
├── templates/
│   └── detailed-feature-docs-template.md  # MASTER TEMPLATE
└── business-features/
    ├── {Module}/                     # One folder per service/module in your project
    │   ├── README.md                 # Complete module documentation
    │   ├── INDEX.md                  # Navigation hub
    │   ├── API-REFERENCE.md          # Endpoint documentation
    │   ├── TROUBLESHOOTING.md        # Issue resolution guide
    │   └── detailed-features/
    │       └── README.{FeatureName}.md     # Comprehensive (17-section, max 1200 lines)
    └── ...

Module Mapping

Search your codebase to discover the module-to-folder mapping:

# Find all service directories
ls -d src/Services/*/

# Find all existing feature doc modules
ls -d docs/business-features/*/

Map each module code to its folder name and service path. Example pattern:

Module CodeFolder NameService Path
{Module1}{Module1}src/Services/{Module1}/
{Module2}{Module2}src/Services/{Module2}/

MANDATORY 17-SECTION STRUCTURE

All feature documentation MUST ATTENTION follow this section order:

#SectionAudience
1Header + Metadata (YAML frontmatter)All
2GlossaryAll
3Executive SummaryPO, BA
4Business RequirementsBA, Dev
5Domain ModelDev, Architect
6Business RulesBA, Dev
7Process FlowsBA, Dev
8Commands & OperationsDev
9Events & Background JobsDev
10UI PagesDev, UX
11API Reference (Simplified)Dev
12Cross-Service IntegrationArchitect
13Security & PermissionsDev, Architect
14Performance ConsiderationsDev, Architect
15Test SpecificationsQA, Dev
16TroubleshootingDev, QA
17Version HistoryAll

Stakeholder Quick Navigation

AudienceSections
Product OwnerExecutive Summary, Business Requirements
Business AnalystBusiness Requirements, Business Rules, Process Flows, Domain Model
DeveloperDomain Model, Commands & Operations, API Reference, Events, UI Pages
Technical ArchitectDomain Model, Cross-Service Integration, Security, Performance
QA/QCTest Specifications, Business Rules, Troubleshooting
UX DesignerUI Pages, Process Flows

Phase 1: Module Detection & Context Gathering

Step 1.0: Auto-Detect Modules from Git Changes (Default)

When no module or feature is explicitly specified, automatically detect affected modules from git changes:

  1. Run git diff --name-only HEAD (captures both staged and unstaged changes)
  2. If no uncommitted changes, run git diff --name-only HEAD~1 (last commit)
  3. Extract unique module names from changed file paths using the Module Mapping table
  4. For each detected module, check if a business feature doc exists in docs/business-features/{Module}/
  5. If docs exist → proceed to Phase 1.5 (Update Mode) for each module
  6. If no docs exist → skip (do not create docs from scratch without explicit user request)
  7. If no service-layer files changed (e.g., only .claude/, docs/, config files) → report "No business feature docs impacted" and exit

Path-to-Module Detection Rules:

Search your codebase to build the path-to-module mapping. Common patterns:

Changed File Path PatternDetected Module
src/Services/{Module}/**{Module}
{frontend-apps-dir}/{app-name}/**{Module} (map app name to module)
{frontend-libs-dir}/{domain-lib}/src/{feature}/**{Module} (map feature to module)

Build a project-specific mapping by examining:

ls -d src/Services/*/
ls -d {frontend-apps-dir}/*/

Step 1.1: Identify Target Module

Determine which module the feature belongs to by:

  1. User explicitly specifies module name
  2. Feature name/domain implies module (search codebase to verify mapping)
  3. Search codebase for feature-related entities/commands
  4. Auto-detected from git diff (Step 1.0 above) — used when invoked as a workflow step without explicit module

Step 1.2: Read Existing Documentation

Before creating new docs, read existing structure:

  1. Read docs/BUSINESS-FEATURES.md (master index)
  2. Read docs/business-features/{Module}/INDEX.md (if exists)
  3. Read docs/business-features/{Module}/README.md (if exists)
  4. Identify what already exists vs what needs creation/update

Step 1.3: Codebase Analysis

Gather evidence from source code:

  • Entities: src/Services/{Module}/{Module}.Domain/Entities/
  • Commands: src/Services/{Module}/{Module}.Application/UseCaseCommands/
  • Queries: src/Services/{Module}/{Module}.Application/UseCaseQueries/
  • Controllers: src/Services/{Module}/{Module}.Service/Controllers/
  • Frontend: {frontend-apps-dir}/{app-name}/ or {frontend-libs-dir}/{domain-lib}/

Step 1.4: Feature Analysis

Build knowledge model in .ai/workspace/analysis/[feature-name].md.

Discovery Areas
  1. Domain Entity Discovery: Entities, value objects, enums
  2. Workflow Discovery: Commands, Queries, Event Handlers, Background Jobs
  3. API Discovery: Controllers, endpoints, DTOs
  4. Frontend Discovery: Components, Services, Stores
  5. Cross-Service Discovery: Message Bus messages, producers, consumers

Phase 1.5: Update Mode (when updating existing docs)

When UPDATING an existing business feature document (not creating from scratch):

Step 1.5.1: Diff Analysis

  1. Identify the source of changes (git diff, branch comparison, commit history)
  2. Categorize changes by type: backend entity, command, query, frontend component, i18n, etc.
  3. Map each change to impacted documentation sections (use table below)

Step 1.5.2: Section Impact Mapping

Change TypeImpacted Sections
New entity property4 (Business Requirements), 5 (Domain Model), 11 (API Reference)
New API endpoint11 (API Reference), 13 (Security & Permissions)
New frontend component10 (UI Pages)
New filter/query4 (Business Requirements), 11 (API Reference)
Any new functionality15 (Test Specifications) — MANDATORY
Any change3 (Executive Summary), 17 (Version History) — ALWAYS UPDATE

Step 1.5.3: Mandatory Test Coverage (Section 15)

CRITICAL: When documenting ANY new functionality, you MUST ATTENTION update:

  • Section 15 (Test Specifications): Add test cases (TC-{FEATURE}-{NNN}) for new features with GIVEN/WHEN/THEN format. Test data, edge cases, and regression impact are included inline within each test case. Each TC entry should include: ` #### TC-GM-001: Create SMART Goal Successfully **Priority**: P0-Critical **Status**: Tested | Untested **Business Rules**: BR-GM-001, BR-GM-003 **IntegrationTest**: Orders/OrderCommandIntegrationTests.cs::SaveOrder_WhenValidData_ShouldCreateSuccessfully **Evidence**: {Service}.Application/{Feature}/Commands/Save{Feature}Command.cs:42-68 **Edge Cases**: - {Invalid scenario} -> {Expected error/behavior} `

Failure to update Section 15 is a blocking quality issue.

Step 1.5.4: CHANGELOG Entry

Always create/update CHANGELOG.md entry under [Unreleased] following Keep a Changelog format.


Phase 2: Documentation Generation

Generate at docs/business-features/{Module}/detailed-features/README.{FeatureName}.md.

Key Format Examples

Business Requirements (FR-XX):

#### FR-{MOD}-01: {Requirement Title}

| Aspect          | Details                           |
| --------------- | --------------------------------- |
| **Description** | {What this requirement enables}   |
| **Scope**       | {Who can use / affected entities} |
| **Evidence**    | `{FilePath}:{LineRange}`          |

User Stories (US-XX):

#### US-{MOD}-01: {Story Title}

**As a** {role}
**I want** {goal/desire}
**So that** {benefit/value}

**Acceptance Criteria**:

- [ ] AC-01: {Criterion with evidence reference}
- [ ] AC-02: {Criterion with evidence reference}

**Related Requirements**: FR-{MOD}-01, FR-{MOD}-02
**Evidence**: `{FilePath}:{LineRange}`

Test Summary Table (MANDATORY):

| Category    | P0 (Critical) | P1 (High) | P2 (Medium) | P3 (Low) |  Total  |
| ----------- | :-----------: | :-------: | :---------: | :------: | :-----: |
| {Category1} |      {N}      |    {N}    |     {N}     |   {N}    |   {N}   |
| **Total**   |    **{N}**    |  **{N}**  |   **{N}**   | **{N}**  | **{N}** |

Test Case Format (TC-XX):

#### TC-{FEATURE}-001: {Test Name} [P0]

**Acceptance Criteria**:

- {Passing criteria 1}
- {Passing criteria 2}

**GIVEN** {initial context}
**WHEN** {action performed}
**THEN** {expected outcome}

**Edge Cases**:

- {Invalid scenario} → {Expected error/behavior}

**Evidence**: `{FilePath}:{LineRange}`

Troubleshooting Format:

#### {Issue Title}

**Symptoms**: {Observable problem}

**Causes**:

1. {Cause 1}
2. {Cause 2}

**Resolution**:

- {Step 1}
- {Step 2}

Permission Matrix:

| Role  | View | Create | Edit | Delete | Special     |
| ----- | :--: | :----: | :--: | :----: | ----------- |
| Admin |  ✅  |   ✅   |  ✅  |   ✅   | Full access |

Note: AI Companion Files Deprecated

As of 2026-04-07, .ai.md companion files are no longer generated. Single README.{Feature}.md is the only output. The 17-section template at docs/templates/detailed-feature-docs-template.md is the authoritative source.

Key Principles (v3.0)

  • No code details in docs -- no file paths, no C# types, no API shapes in sections 1-14, 16
  • Evidence only in Section 15 (Test Specifications) -- file:line references
  • Commands cross-reference BR-XXX -- each command lists which business rules it validates
  • Max 1200 lines per doc (target 500-800)
  • YAML frontmatter required: module, service, feature_code, entities[], status, last_updated

Phase 3: Master Index Update

After creating/updating module docs, update docs/BUSINESS-FEATURES.md:

  1. Read current content
  2. Verify module is listed in the "Detailed Module Documentation" table
  3. Add link if missing: | **{Module}** | [Description] | [View Details](./business-features/{Module}/README.md) |

Anti-Hallucination Protocols

Evidence-Based Reasoning — Speculation is FORBIDDEN. Every claim needs proof. 1. Cite file:line, grep results, or framework docs for EVERY claim 2. Declare confidence: >80% act freely, 60-80% verify first, <60% DO NOT recommend 3. Cross-service validation required for architectural changes 4. "I don't have enough evidence" is valid and expected output BLOCKED until: - [] Evidence file path (file:line) - [] Grep search performed - [] 3+ similar patterns found - [] Confidence level stated Forbidden without proof: "obviously", "I think", "should be", "probably", "this is because" If incomplete → output: "Insufficient evidence. Verified: [...]. Not verified: [...]."

DOCUMENTATION_ACCURACY_CHECKPOINT

Before writing any documentation:

  • "Have I read the actual code that implements this?"
  • "Are my line number references accurate and current?"
  • "Can I provide a code snippet as evidence?"

TEST CASE EVIDENCE VERIFICATION

For EVERY test case:

  1. Read the Evidence file at claimed line number
  2. Verify: Does code at that line support test assertion?
  3. Check Edge Cases: Find error constants in ErrorMessage.cs
  4. Fix immediately if line numbers wrong

Phase 3.5: Verification (3 Passes)

First Pass - Test Case Evidence Audit (Section 15)

For EVERY test case in documentation:

  1. Read the Evidence file at the claimed line number
  2. Verify match: Does the code at that line support the test assertion?
  3. Check Edge Cases: Find error message constants in ErrorMessage.cs
  4. Fix immediately if line numbers are wrong

Second Pass - Domain Model Verification

  • Read EACH entity file referenced in Domain Model section (Section 5)
  • Verify property names and business meanings are accurate (no C# types -- use business meaning column)
  • Check enum values exist in actual source
  • Remove any documented properties not found in source

Third Pass - Cross-Reference Audit

  • Document has exactly 17 sections in correct order
  • Test Summary counts match actual test case count in Section 15
  • All internal links work
  • No template placeholders remain ({FilePath}, {LineRange})
  • ErrorMessage.cs constants match edge case messages
  • YAML frontmatter is present and complete

CRITICAL: If ANY pass finds hallucinated content, re-investigate and fix before completing.


Quality Checklist

Structure

  • Documentation placed in correct folder structure
  • README.md follows template format (17 sections)
  • YAML frontmatter present with module, service, feature_code, entities[]
  • INDEX.md created with navigation links
  • Master index (BUSINESS-FEATURES.md) updated
  • Stakeholder navigation table present
  • CHANGELOG.md updated with entry under [Unreleased]
  • Max 1200 lines total document length
  • No code details in sections 1-14, 16 (no file paths, no C# types)
  • Commands reference BR-XXX IDs they validate

Test Case Evidence (MANDATORY)

  • EVERY test case has Evidence field with file:line format
  • No template placeholders remain ({FilePath}, {LineRange})
  • Line numbers verified by reading actual source files
  • Edge case errors match constants from ErrorMessage.cs
  • Test Summary counts match actual number of test cases in Section 15

Anti-Hallucination

  • All entity properties verified against source code
  • All enum values verified against actual enum definitions
  • No invented methods, properties, or models
  • All code snippets copied from actual files

Related

  • documentation
  • feature-implementation

Next Steps

MANDATORY IMPORTANT MUST ATTENTION — NO EXCEPTIONS after completing this skill, you MUST ATTENTION use AskUserQuestion to present these options. Do NOT skip because the task seems "simple" or "obvious" — the user decides:

  • "/tdd-spec (Recommended)" — Generate/update test specs for documented features
  • "/test-specs-docs" — Sync test specs to dashboard
  • "Skip, continue manually" — user decides

Closing Reminders

  • IMPORTANT MUST ATTENTION break work into small todo tasks using TaskCreate BEFORE starting
  • IMPORTANT MUST ATTENTION search codebase for 3+ similar patterns before creating new code
  • IMPORTANT MUST ATTENTION cite file:line evidence for every claim (confidence >80% to act)
  • IMPORTANT MUST ATTENTION add a final review todo task to verify work quality MANDATORY IMPORTANT MUST ATTENTION READ the following files before starting:
  • IMPORTANT MUST ATTENTION cite file:line evidence for every claim. Confidence >80% to act, <60% = do NOT recommend.

平台分布

Claude Code

26.48%
按下载量换算79

windsurf

23.84%
按下载量换算71

OpenCode

19.47%
按下载量换算58

Codex

12.51%
按下载量换算37

Antigravity

7.6%
按下载量换算23

Gemini CLI

3.45%
按下载量换算10

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。

继续浏览同类 Skills