@tensorify.io/cli

Official CLI for Tensorify.io - Build, test, and deploy machine learning plugins

The Tensorify CLI provides powerful tools for developing, validating, and publishing machine learning plugins to the Tensorify ecosystem.

🚀 Quick Start

Installation

# Install globally
npm install -g @tensorify.io/cli

# Or install locally in your project
npm install --save-dev @tensorify.io/cli

Authentication

# Login to your Tensorify account
tensorify login

# Check your authentication status
tensorify whoami

# Logout when needed
tensorify logout

Basic Usage

# Validate your plugin structure
tensorify validate

# Publish your plugin
tensorify publish --access=public

📋 Commands

tensorify login

Authenticate with Tensorify.io using OAuth2 flow.

tensorify login [options]

Features:

• 🔐 Secure OAuth2 authentication
• 🌐 Opens browser for login
• 💾 Stores token securely in system keyring
• ✅ Automatic token validation

Options:

• --port <number> - Custom port for OAuth callback (default: 3000)
• --timeout <seconds> - Authentication timeout (default: 300)

Examples:

# Standard login
tensorify login

# Login with custom port
tensorify login --port 8080

# Login with timeout
tensorify login --timeout 600

tensorify whoami

Display current authentication status and user information.

tensorify whoami [options]

Features:

• 👤 Shows logged-in user details
• 🔑 Displays token status
• ⏰ Shows token expiration
• 📊 User account information

Options:

• --json - Output in JSON format
• --token - Show token details

Examples:

# Basic user info
tensorify whoami

# JSON output
tensorify whoami --json

# Include token details
tensorify whoami --token

tensorify logout

Clear authentication tokens and logout.

tensorify logout [options]

Features:

• 🗑️ Removes stored tokens
• 🔒 Secure cleanup
• ✅ Confirmation prompts

Options:

• --force - Skip confirmation prompt

Examples:

# Standard logout
tensorify logout

# Force logout without confirmation
tensorify logout --force

tensorify validate

Validate plugin structure and configuration using SDK rules.

tensorify validate [directory] [options]

Features:

• 🔍 Complete plugin validation
• 📁 File structure checks
• 📝 Schema validation
• 🏗️ Class interface validation
• 📊 Detailed error reporting

Arguments:

• directory - Plugin directory path (default: current directory)

Options:

• --verbose - Show detailed validation output
• --sdk-version <version> - Check against specific SDK version
• --json - Output results in JSON format
• --fix - Attempt to fix common issues

Validation Checks:

• ✅ Required files exist (index.ts, manifest.json, icon.svg, package.json)
• ✅ manifest.json schema validation
• ✅ package.json structure validation
• ✅ Class implements INode interface
• ✅ Class name matches manifest entrypointClassName
• ✅ SDK version compatibility
• ✅ Repository URL for public plugins

Examples:

# Validate current directory
tensorify validate

# Validate specific directory
tensorify validate ./my-plugin

# Verbose validation output
tensorify validate --verbose

# JSON output for CI/CD
tensorify validate --json

# Check against specific SDK version
tensorify validate --sdk-version 0.0.1

tensorify publish

Publish plugin to Tensorify registry with comprehensive validation and upload.

tensorify publish [options]

Features:

• 🔍 Pre-publish validation
• 🏗️ Automatic build and bundling
• ✅ Version conflict checking
• 📤 Secure file uploads
• 🔔 Registry notifications
• 🚀 Complete publishing pipeline

Options:

• --access <level> - Access level: public or private (default: public)
• --directory <path> - Plugin directory (default: current directory)
• --backend <url> - Backend API URL (default: https://backend.tensorify.io)
• --frontend <url> - Frontend API URL (default: https://plugins.tensorify.io)
• --dry-run - Validate and build without publishing
• --skip-build - Skip build step (use existing dist/)
• --force - Force publish (skip confirmations)
• --offline - Offline development mode (implies --dev): save artifacts to local offline-plugins/ dir and notify backend without S3 upload

Publishing Process:

1. 🔐 Authentication Check

   • Verifies valid login token
   • Checks token expiration

2. 🔍 Plugin Validation

   • Runs complete SDK validation
   • Checks file structure
   • Validates schemas

3. 🔒 Access Level Validation

   • Ensures package.json private flag matches --access
   • Validates repository URL for public plugins
   • Checks access consistency with previous versions

4. ✅ Version Conflict Check

   • Queries registry for existing versions
   • Prevents duplicate version publishing
   • Validates access level consistency

5. 🏗️ Build and Bundle

   • Runs TypeScript compilation (pnpm run build)
   • Creates optimized bundle with ESBuild
   • Generates production-ready artifacts

6. 📤 File Upload

   • Uploads bundle.js, manifest.json, icon.svg (online)
   • OR saves them to local offline-plugins/@<username>/<plugin>:<version>/ (offline)
   • Provides upload/save progress

7. 🔔 Registry Notification

   • Sends webhook to registry
   • Updates plugin database
   • Triggers indexing

Examples:

# Publish public plugin
tensorify publish --access=public

# Publish private plugin
tensorify publish --access=private

# Dry run (validate without publishing)
tensorify publish --dry-run

# Publish from specific directory
tensorify publish --directory ./my-plugin

# Force publish without confirmations
tensorify publish --force

# Custom backend/frontend URLs
tensorify publish --backend https://api.custom.com --frontend https://registry.custom.com

# Offline development (no S3 upload, implies --dev)
tensorify publish --offline

🔧 Configuration

Environment Variables

# Backend API URL
export TENSORIFY_BACKEND_URL=https://backend.tensorify.io

# Frontend registry URL
export TENSORIFY_FRONTEND_URL=https://plugins.tensorify.io

# OAuth configuration
export TENSORIFY_CLIENT_ID=your_client_id
export TENSORIFY_CLIENT_SECRET=your_client_secret

# CLI configuration
export TENSORIFY_DEFAULT_ACCESS=public
export TENSORIFY_CLI_LOG_LEVEL=info

Configuration File

Create .tensorifyrc.json in your project or home directory:

{
  "backend": "https://backend.tensorify.io",
  "frontend": "https://plugins.tensorify.io",
  "defaultAccess": "public",
  "autoValidate": true,
  "buildTimeout": 300,
  "uploadTimeout": 600
}

🔐 Authentication

OAuth2 Flow

The CLI uses secure OAuth2 authentication:

1. Login Command: Opens browser to Tensorify.io
2. User Authorization: User grants CLI access
3. Token Exchange: CLI receives secure access token
4. Token Storage: Token stored in system keyring
5. Automatic Refresh: Tokens refreshed as needed

Token Storage

Tokens are securely stored using:

• macOS: Keychain
• Windows: Credential Manager
• Linux: Secret Service (libsecret)

Token Security

• 🔒 Tokens encrypted at rest
• ⏰ Automatic expiration
• 🔄 Refresh token rotation
• 🚫 No tokens in environment variables

📦 Plugin Requirements

Required Files

my-plugin/
├── package.json      # NPM package configuration
├── manifest.json     # Plugin metadata
├── icon.svg         # Plugin icon (SVG format)
├── index.ts         # Main plugin class
├── tsconfig.json    # TypeScript configuration
└── dist/           # Build output directory

package.json Structure

{
  "name": "@namespace/plugin-name",
  "version": "1.0.0",
  "main": "dist/index.js",
  "scripts": {
    "build": "tsc"
  },
  "tensorify-settings": {
    "sdk-version": "0.0.1"
  },
  "keywords": ["tensorify", "plugin"],
  "author": "Your Name",
  "repository": {
    "type": "git",
    "url": "https://github.com/username/plugin-name"
  },
  "private": false
}

manifest.json Structure

{
  "name": "@namespace/plugin-name",
  "version": "1.0.0",
  "description": "Plugin description",
  "author": "Your Name",
  "main": "dist/index.js",
  "entrypointClassName": "MyPluginClass",
  "keywords": ["tensorify", "plugin"],
  "scripts": {
    "build": "tsc"
  },
  "tensorifySettings": {
    "sdkVersion": "0.0.1"
  }
}

Plugin Class Structure

import { INode, NodeType, LayerSettings } from "@tensorify.io/sdk";

export default class MyPluginClass implements INode {
  readonly name = "My Plugin";
  readonly nodeType = NodeType.CUSTOM;
  readonly inputLines = 1;
  readonly outputLinesCount = 1;
  readonly secondaryInputLinesCount = 0;
  readonly translationTemplate = "my_template";
  readonly settings: LayerSettings = {};

  getTranslationCode(settings: LayerSettings): string {
    return "# Generated code";
  }

  validateSettings(settings: LayerSettings): boolean {
    return true;
  }

  getDependencies(): string[] {
    return ["numpy", "torch"];
  }

  getImports(): string[] {
    return ["import torch"];
  }
}

🔄 CI/CD Integration

GitHub Actions

name: Publish Plugin

on:
  push:
    tags: ["v*"]

jobs:
  publish:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v3

      - name: Setup Node.js
        uses: actions/setup-node@v3
        with:
          node-version: "18"

      - name: Install dependencies
        run: npm install

      - name: Install Tensorify CLI
        run: npm install -g @tensorify.io/cli

      - name: Authenticate
        run: tensorify login --token ${{ secrets.TENSORIFY_TOKEN }}

      - name: Validate plugin
        run: tensorify validate --json

      - name: Publish plugin
        run: tensorify publish --access=public

GitLab CI

stages:
  - validate
  - publish

validate:
  stage: validate
  script:
    - npm install -g @tensorify.io/cli
    - tensorify validate --json
  only:
    - merge_requests

publish:
  stage: publish
  script:
    - npm install -g @tensorify.io/cli
    - echo $TENSORIFY_TOKEN | tensorify login --stdin
    - tensorify publish --access=public
  only:
    - tags

🐛 Troubleshooting

Common Issues

Authentication Problems

❌ "Not authenticated"

# Solution: Login first
tensorify login

# Check authentication status
tensorify whoami

❌ "Token expired"

# Solution: Re-login
tensorify logout
tensorify login

Validation Errors

❌ "Plugin validation failed"

# Get detailed validation info
tensorify validate --verbose

# Common fixes:
# 1. Check file names (index.ts, manifest.json, icon.svg)
# 2. Verify class name matches manifest entrypointClassName
# 3. Ensure class implements INode interface
# 4. Check SDK version compatibility

❌ "SDK version mismatch"

// Update package.json
{
  "tensorify-settings": {
    "sdk-version": "0.0.1" // Use current SDK version
  }
}

❌ "Missing repository URL"

// For public plugins, add repository
{
  "repository": {
    "type": "git",
    "url": "https://github.com/username/plugin-name"
  },
  "private": false
}

Publishing Problems

❌ "Version already exists"

# Solution: Increment version in package.json and manifest.json
# Cannot republish existing versions

❌ "Access level mismatch"

# Solution: Use consistent access level
# Cannot change from public to private or vice versa

❌ "Build failed"

# Check TypeScript compilation
pnpm run build

# Verify tsconfig.json exists
# Ensure all dependencies installed

❌ "Upload failed"

# Check network connection
# Verify file sizes (< 100MB each)
# Ensure backend is accessible

Debug Mode

Enable detailed logging:

# Set log level
export TENSORIFY_CLI_LOG_LEVEL=debug

# Or use debug flag
tensorify validate --debug
tensorify publish --debug

Getting Help

# Command help
tensorify --help
tensorify publish --help

# Version information
tensorify --version

# Validate with verbose output
tensorify validate --verbose

📈 Performance Tips

Faster Builds

// tsconfig.json optimization
{
  "compilerOptions": {
    "incremental": true,
    "tsBuildInfoFile": ".tsbuildinfo"
  }
}

Faster Uploads

• Keep bundle sizes small (< 10MB recommended)
• Optimize dependencies
• Use .npmignore to exclude unnecessary files

Faster Validation

# Skip build for validation-only
tensorify validate --skip-build

# Use specific directory
tensorify validate ./plugin-dir

🔗 Integration Examples

Pre-commit Hooks

// package.json
{
  "husky": {
    "hooks": {
      "pre-commit": "tensorify validate"
    }
  }
}

NPM Scripts

// package.json
{
  "scripts": {
    "validate": "tensorify validate",
    "publish:public": "tensorify publish --access=public",
    "publish:private": "tensorify publish --access=private",
    "publish:dry": "tensorify publish --dry-run"
  }
}

📚 Resources

• 📖 Documentation: docs.tensorify.io
• 🎓 Plugin Development Guide: docs.tensorify.io/plugins
• 🔧 SDK Documentation: @tensorify.io/sdk
• 💬 Community Discord: discord.gg/tensorify
• 🐛 Issue Tracker: github.com/tensorify/cli/issues
• 📧 Support: [e-mail]

📄 License

MIT License - see LICENSE file for details.

---

Made with ❤️ by the Tensorify Team