technical-doc-generator（技术文档生成器）

Technical Documentation Generator

Scan a codebase and generate client-deliverable technical documentation. Produces API docs, READMEs, architecture overviews, changelogs, and developer onboarding guides.

How to Use

/technical-doc-generator ./src api-docs
/technical-doc-generator . full
/technical-doc-generator ./src readme
/technical-doc-generator . changelog
/technical-doc-generator ./src onboarding

• $ARGUMENTS[0] = Path to codebase (default: current directory)
• $ARGUMENTS[1] = Doc type: api-docs, readme, architecture, changelog, onboarding, full (all)

Documentation Types

readme — Project README

Scan the project and generate a comprehensive README:

# Project Name

Brief description (1-2 sentences from package.json, pyproject.toml, or code analysis).

## Features
- [Auto-detected from code structure]

## Quick Start

### Prerequisites
[Detected from package.json, requirements.txt, go.mod, etc.]

### Installation
[Generated from detected package manager and config files]

### Configuration
[Detected from .env.example, config files, environment variables in code]

### Usage
[Basic usage examples from entry points, CLI args, or main functions]

## Project Structure
[Generated directory tree with descriptions]

## API Reference
[Brief overview with link to full docs if generated]

## Contributing
[Standard contributing section]

## License
[Detected from LICENSE file]

api-docs — API Documentation

Scan for API endpoints and generate documentation:

1. Detect the framework: Express, FastAPI, Django, Flask, Rails, Spring, Gin, etc.
2. Extract endpoints: Routes, methods, parameters, request/response bodies
3. Generate OpenAPI/Swagger spec (YAML)
4. Generate human-readable docs (Markdown)

For each endpoint:

### `POST /api/users`

Create a new user account.

**Authentication**: Required (Bearer token)

**Request Body**:
| Field | Type | Required | Description |
|-------|------|----------|-------------|
| email | string | Yes | User's email address |
| name | string | Yes | Full name |
| role | string | No | User role (default: "member") |

**Example Request**:

{ "email": "user@example.com", "name": "Jane Doe", "role": "admin" }

**Response** (`201 Created`):

{ "id": "usr_abc123", "email": "user@example.com", "name": "Jane Doe", "role": "admin", "created_at": "2026-02-13T10:00:00Z" }

**Error Responses**:
| Status | Description |
|--------|-------------|
| 400 | Invalid request body |
| 409 | Email already exists |
| 401 | Missing or invalid auth token |

architecture — Architecture Overview

Generate an architecture document with:

1. System overview: What the system does, high-level description
2. Technology stack: Languages, frameworks, databases, services detected
3. Directory structure: Annotated tree with purpose of each directory
4. Component diagram: Mermaid diagram showing major components and their relationships
5. Data flow: How data moves through the system
6. Database schema: If migrations or models are found, document the schema
7. External dependencies: Third-party services, APIs, databases
8. Configuration: Environment variables and their purposes

Mermaid diagram example:

graph TB
    Client[Client App] --> API[API Server]
    API --> Auth[Auth Service]
    API --> DB[(PostgreSQL)]
    API --> Cache[(Redis)]
    API --> Queue[Job Queue]
    Queue --> Worker[Background Worker]
    Worker --> DB
    Worker --> Email[Email Service]

changelog — Changelog from Git History

Parse git log and generate a structured changelog:

# Changelog

## [Unreleased]
### Added
- [Features from commits since last tag]

### Changed
- [Modifications]

### Fixed
- [Bug fixes]

## [1.2.0] - 2026-02-10
### Added
- ...

Rules:

• Group by semantic version tags (or by month if no tags)
• Categorize commits: Added, Changed, Fixed, Removed, Security, Deprecated
• Parse conventional commits if used (feat:, fix:, chore:, etc.)
• Skip merge commits and build/CI commits
• Link to PRs/issues if references found in commit messages

onboarding — Developer Onboarding Guide

Generate a guide for new developers joining the project:

# Developer Onboarding Guide

## Prerequisites
[Required software, versions, accounts]

## Getting Started
### 1. Clone and Setup
[Step-by-step with exact commands]

### 2. Environment Configuration
[All env vars explained with example values]

### 3. Run Locally
[Commands to start the dev server, run tests, etc.]

### 4. Verify Setup
[How to confirm everything is working]

## Codebase Tour
### Architecture Overview
[Brief system description with diagram]

### Key Directories
[What lives where and why]

### Important Files
[Config files, entry points, key modules]

## Development Workflow
### Branching Strategy
[Detected from git history or standard gitflow]

### Running Tests
[Test commands, test structure]

### Code Style
[Linting config, formatting tools detected]

### Making Changes
[Typical workflow: branch → code → test → PR]

## Common Tasks
### Add a new API endpoint
[Step-by-step based on existing patterns]

### Add a database migration
[Based on detected ORM/migration tool]

### Deploy
[If deployment config is detected]

## Troubleshooting
### Common Issues
[Based on README, issues, or common patterns]

full — Complete Documentation Package

Generate ALL of the above, organized in a docs/ directory:

docs/
  README.md              # Project README (also copy to project root if none exists)
  API.md                 # API documentation
  ARCHITECTURE.md        # Architecture overview
  CHANGELOG.md           # Changelog
  ONBOARDING.md          # Developer onboarding
  openapi.yaml           # OpenAPI spec (if API detected)
  diagrams/              # Mermaid source files

Codebase Scanning Strategy

1. Package files first: package.json, pyproject.toml, go.mod, Cargo.toml, Gemfile, pom.xml
2. Config files: .env.example, docker-compose.yml, CI/CD configs
3. Entry points: main.*, index.*, app.*, server.*
4. Route/endpoint files: Files containing route definitions
5. Models/schemas: Database models, TypeScript interfaces, Pydantic models
6. Test files: To understand expected behavior
7. Existing docs: Any existing README, docs/, wiki content

Do NOT read every file. Be strategic — scan structure first, then dive into key files.

Output

Save all generated docs to output/docs/ (or docs/ if specified). Present a summary of what was generated and suggest next steps for the client.