DocuMind

AI-Powered Swagger/OpenAPI Documentation Generator for Express.js projects with intelligent context stitching.

🚀 Why DocuMind?

DocuMind is designed to fix the common problem of shallow or empty generated API docs. Instead of only parsing route definitions, it intelligently collects surrounding business logic so Gemini understands each endpoint in context.

• Extracts routes, controllers, services, models, and validations
• Encourages deep summaries and descriptions from Gemini
• Prevents empty paths and weak documentation
• Keeps generated output aligned with your implementation

🎯 Showcase

Input

A simple Express route plus related service and validation code.

// src/routes/auth.ts
router.post('/login', authController.login);

Output

openapi: 3.0.0
info:
  title: DocuMind Generated API
  version: 1.0.0
  description: Automated OpenAPI documentation generated by DocuMind AI.
paths:
  /login:
    post:
      tags:
        - Auth
      summary: Authenticate user and issue a JWT token.
      description: |
        Verifies credentials against the hashed user password, checks authorization
        roles, signs a JWT token, and returns a session payload.
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/LoginRequest'
      responses:
        '200':
          description: User authenticated successfully.
components:
  schemas:
    LoginRequest:
      type: object
      properties:
        email:
          type: string
          format: email
        password:
          type: string

🧠 How it Works

Context Stitching

DocuMind attaches nearby compute context to route files so Gemini sees the full backend flow.

flowchart LR
  A[Route File] --> B[Controller Code]
  A --> C[Service Logic]
  A --> D[Validation / Models]
  B --> E[AI Prompt]
  C --> E
  D --> E
  E --> F[Gemini AI]
  F --> G[OpenAPI Output]

Generation Flow

flowchart TD
  A[CLI Start] --> B[Parse CLI Options]
  B --> C[Scan Project Directory]
  C --> D[Filter Express Route Files]
  D --> E[Collect Related Context]
  E --> F[Call Gemini Model]
  F --> G[Defensive Merge Output]
  G --> H[Write swagger.yaml / json]

✨ Key Features

• Smart File Scanning: finds Express route files across the project.
• Context Stitching: includes controllers, services, models, and validation code.
• Defensive Merging: merges AI-generated paths and schemas safely.
• Rich Descriptions: forces Gemini to explain workflows like hashing, JWT creation, roles, and errors.
• Interactive CLI: animated progress using ora.

🧰 Tech Stack

• Node.js (>= 18.0.0)
• TypeScript
• Google Gen AI SDK (@google/genai)
• Commander.js
• Ora
• Dotenv
• yaml

📥 Installation

Global

npm install -g @mena-emad/documind

Local

git clone https://github.com/mena-emad/DocuMind.git
cd DocuMind
npm install
npm run build

Run in development mode:

npm run dev

Run locally:

npx @mena-emad/documind

🔧 Environment Variables

DocuMind reads the Gemini API key from .env:

DOCUMIND_API_KEY=your_gemini_api_key_here

Or pass the key directly using -k.

▶️ Usage

documind [options]

CLI Options

| Option | Description | Default | |---|---|---| | -k, --key <api-key> | Gemini API Key | DOCUMIND_API_KEY from .env | | -d, --dir [directory] | Directory to scan | ./ | | -o, --output [file-name] | Output filename | swagger.yaml | | --title [title] | OpenAPI info.title | auto-generated | | --api-version [version] | OpenAPI info.version | auto-generated | | --desc [description] | OpenAPI info.description | auto-generated | | -v, --version | Show CLI version | - |

Examples

Generate default output:

documind

Generate docs for a custom source directory:

documind -d ./src -o api-docs.yaml

Use the API key directly:

documind -k sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Provide OpenAPI metadata:

documind --title "My App API" --api-version "2.0.0" --desc "AI-generated Swagger docs from Express source code."

🔍 How DocuMind Processes Code

1. Scan files
   • Recursively collect .js and .ts files.
2. Filter Express routes
   • Detect route patterns and Express router usage.
3. Gather related context
   • Add matching service, controller, model, and validation files.
4. Ask Gemini
   • Send a prompt that demands valid OpenAPI JSON and rich documentation.
5. Merge outputs defensively
   • Combine returned paths and components while handling missing or malformed sections.
6. Write output
   • Save YAML or JSON based on the requested filename.

👨‍💻 About the Author

Built by Mena Emad Sawares.

• Website: https://menaemad.vercel.app/
• GitHub: https://github.com/mena-emad
• Email: [email protected]

📜 License

DocuMind is released under the MIT License.