VibeUniv Getting Started Guide

Follow along step by step to connect your project and start AI-powered learning. No worries if you're new!

Before You Start Sign Up & Log In Register LLM API Key (Optional)Generate VibeUniv API Key Connect MCP Server Sync Your Project Tech Stack Analysis Start Learning!Connect via REST API Without MCP Troubleshooting Security

Before You Start

You just need these 3 things to get started.

AI Coding Tool — Claude Code, Cursor, Windsurf, Cline, Gemini CLI, Kimi Code, or OpenAI Codex installed
An AI-Built Project — The project folder you want to analyze should be on your local machine
LLM API Key — API key for Anthropic, OpenAI, Google, etc. (for Pro plan BYOK; not needed for Free plan)

Sign Up & Log In

Register LLM API Key (Optional)

To use BYOK (Bring Your Own Key) on Pro plan or above, register your LLM API key. If you're on the Free plan, you can skip this step.

1. After logging in, go to Settings > LLM Keys
2. Select your provider (Anthropic, OpenAI, etc.)
3. Enter your API key and save

Your keys are encrypted with AES-256-GCM before storage. VibeUniv never stores keys in plaintext or sends them externally.

Generate VibeUniv API Key

The MCP server needs a VibeUniv API key to link projects to your account.

1. Go to Settings > API Keys
2. Click "Create new key"
3. Copy the generated key and store it somewhere safe

The API key is shown only once after creation. Make sure to copy it! If lost, you'll need to generate a new one.

Connect MCP Server

Add the MCP server configuration to your AI coding tool. The easiest way is to ask the AI directly!

Claude Code

Tell the AI:

Add an MCP server called vibeuniv. Use npx (not npm install). The command is npx -y @vibeuniv/mcp-server@latest with env var VIBEUNIV_API_KEY set to YOUR_API_KEY_HERE.

Manual setup:

Config file: ~/.claude.json

{
  "mcpServers": {
    "vibeuniv": {
      "command": "npx",
      "args": ["-y", "@vibeuniv/mcp-server@latest"],
      "env": {
        "VIBEUNIV_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}

Or add via CLI command:

claude mcp add vibeuniv -s user -e VIBEUNIV_API_KEY=YOUR_API_KEY_HERE -- npx -y @vibeuniv/mcp-server@latest

Cursor

Tell the AI:

Add vibeuniv MCP server to mcp.json. Use npx (not npm install). Command is npx -y @vibeuniv/mcp-server@latest, with env var VIBEUNIV_API_KEY set to YOUR_API_KEY_HERE.

Manual setup:

Config file: ~/.cursor/mcp.json

{
  "mcpServers": {
    "vibeuniv": {
      "command": "npx",
      "args": ["-y", "@vibeuniv/mcp-server@latest"],
      "env": {
        "VIBEUNIV_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}

Windsurf

Tell the AI:

Add vibeuniv to MCP config. Use npx (not npm install). Command is npx -y @vibeuniv/mcp-server@latest, with env var VIBEUNIV_API_KEY set to YOUR_API_KEY_HERE.

Manual setup:

Config file: ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "vibeuniv": {
      "command": "npx",
      "args": ["-y", "@vibeuniv/mcp-server@latest"],
      "env": {
        "VIBEUNIV_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}

Cline (VS Code)

Tell the AI:

Add vibeuniv MCP server. Use npx (not npm install). Command is npx -y @vibeuniv/mcp-server@latest, with env var VIBEUNIV_API_KEY set to YOUR_API_KEY_HERE.

Manual setup:

In VS Code, open Cline > MCP Servers > Edit Config and add the following.

{
  "mcpServers": {
    "vibeuniv": {
      "command": "npx",
      "args": ["-y", "@vibeuniv/mcp-server@latest"],
      "env": {
        "VIBEUNIV_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}

Kimi Code CLI

Tell the AI:

Add an MCP server called vibeuniv. Use npx (not npm install). Command is npx -y @vibeuniv/mcp-server@latest, with env var VIBEUNIV_API_KEY set to YOUR_API_KEY_HERE.

Manual setup:

Config file: ~/.kimi/mcp.json

{
  "mcpServers": {
    "vibeuniv": {
      "command": "npx",
      "args": ["-y", "@vibeuniv/mcp-server@latest"],
      "env": {
        "VIBEUNIV_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}

Or add directly via CLI: manage MCP servers with kimi mcp and check connection status with /mcp.

Gemini CLI

Tell the AI:

Add an MCP server called vibeuniv. Use npx (not npm install). Command is npx -y @vibeuniv/mcp-server@latest, with env var VIBEUNIV_API_KEY set to YOUR_API_KEY_HERE.

Manual setup:

Config file: ~/.gemini/settings.json

{
  "mcpServers": {
    "vibeuniv": {
      "command": "npx",
      "args": ["-y", "@vibeuniv/mcp-server@latest"],
      "env": {
        "VIBEUNIV_API_KEY": "YOUR_API_KEY_HERE"
      }
    }
  }
}

OpenAI Codex CLI

Tell the AI:

Add vibeuniv MCP server. Use npx (not npm install). Package is @vibeuniv/mcp-server@latest, with env var VIBEUNIV_API_KEY set to YOUR_API_KEY_HERE.

Manual setup:

Config file: ~/.codex/config.toml (TOML format)

[mcp_servers.vibeuniv]
command = "npx"
args = ["-y", "@vibeuniv/mcp-server@latest"]

[mcp_servers.vibeuniv.env]
VIBEUNIV_API_KEY = "YOUR_API_KEY_HERE"

Or add with a single CLI command:

codex mcp add vibeuniv --env VIBEUNIV_API_KEY=YOUR_API_KEY_HERE -- npx -y @vibeuniv/mcp-server@latest

For project-specific settings, add to .codex/config.toml (project root). Both CLI and IDE extension share the same config file.

npx vs npm — Why npx?

Always use npx for MCP server configuration.

npx — Runs the package without installing it. Thanks to the @latest tag, it always uses the newest version with zero maintenance.
npm install -g — Installs the package globally on your machine. Once installed, the version is locked and you must manually update to get new features or bug fixes.
npx automatically downloads and runs the latest server every time, so you always have the newest features without worrying about version management.

If npx doesn't work — npm global install

Only use this method if npx fails due to corporate networks, firewalls, etc.

1. Install globally:

npm install -g @vibeuniv/mcp-server@latest

2. Change command and args in your config file:

{
  "command": "vibeuniv-mcp",
  "args": [],
  "env": {
    "VIBEUNIV_API_KEY": "YOUR_API_KEY_HERE"
  }
}

3. When you need to update, just run the install command again.

Config file paths may change with tool updates. Asking the AI ensures it uses the correct path automatically.

Sync Your Project

Once you've saved the MCP config, follow these steps.

1. Fully quit and restart your AI coding tool
2. In your project folder, ask the AI to call vibeuniv_sync_project
3. Check the VibeUniv dashboard to confirm the project is registered

Example prompts to tell the AI

“Sync my project to VibeUniv”

Project registration + file upload

“Analyze the tech stack of this project”

Tech stack analysis (after sync)

“Generate a beginner learning curriculum for me”

Personalized learning roadmap

“Explain what this code does”

Ask the AI Tutor

With MCP connected, the AI automatically calls VibeUniv tools. No need to memorize any commands!

Sync only transmits config files and source code structure. Sensitive files like .env and certificates are automatically excluded.

Tech Stack Analysis

Once your project is registered, AI can analyze your tech stack.

Method 1: Open your project in the dashboard and click the "Analyze" button
Method 2: Call the vibeuniv_analyze MCP tool from your AI tool

Once analysis is complete, you can see frameworks, libraries, and architecture patterns at a glance.

Start Learning!

After analysis, the really fun part begins.

1. Go to the Learning tab of your project
2. Review the AI-generated personalized learning roadmap
3. Start learning module by module based on your code
4. Ask the AI Tutor any question

Not abstract tutorials — your own code becomes your textbook.

Connect via REST API Without MCP

If you can't use MCP, you can send your project directly via REST API. All you need is an API key.

1. Create Project

curl -X POST https://vibeuniv.com/api/v1/projects \
  -H "Authorization: Bearer YOUR_VIBEUNIV_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "my-awesome-app"}'

2. Upload Files

# Use the project ID from the response
curl -X POST https://vibeuniv.com/api/v1/projects/{project_id}/files \
  -H "Authorization: Bearer YOUR_VIBEUNIV_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "files": [
      {"name": "package.json", "content": "..."},
      {"name": "app/page.tsx", "content": "..."}
    ]
  }'

3. Request Tech Stack Analysis

curl -X POST https://vibeuniv.com/api/v1/projects/{project_id}/analyze \
  -H "Authorization: Bearer YOUR_VIBEUNIV_API_KEY"

Once analysis is complete, you can view results in the dashboard and generate a learning roadmap.

Troubleshooting

MCP server won't connect

Check that the config file JSON format is valid (commas, braces, etc.)
Make sure the VIBEUNIV_API_KEY value is correct
Check that Node.js is installed (npx is required)
Fully quit and restart your AI coding tool

Project doesn't appear in dashboard

Try calling vibeuniv_sync_project again
Make sure the API key belongs to the correct account
Refresh the dashboard in your browser

Analysis fails

Check that the project has dependency files like package.json or requirements.txt
If using BYOK, verify that the registered LLM API key is valid
Try again later — it might be a temporary LLM provider error

I was using VIBESTACK_API_KEY before

The old VIBESTACK_API_KEY still works (backwards compatible)
We recommend switching to VIBEUNIV_API_KEY when you have time

Security

Files We Collect

package.json, requirements.txt and other dependency files
tsconfig.json, next.config.ts and other config files
CLAUDE.md, .cursorrules and other AI project files
Source code (.ts, .tsx, .py, etc. — up to 50 files)

Automatically Excluded Files

.env, .env.* — environment variable files
*.pem, *.key, id_rsa* — certificate/key files
credentials.json, secrets.* — credential files
node_modules/, .git/, dist/ — build/dependencies

All data is transmitted via HTTPS and LLM API keys are encrypted with AES-256-GCM. You can completely delete your project data at any time.

Feature Guide by Menu

Explore detailed usage for Dashboard, Projects, Learning, Settings, and Rewards.

VibeUniv Technology

Discover our ontology-based Knowledge Graph, adaptive curriculum, and more.

Ready to Go?

Connect your project and start learning from your own code.

Get Started