Installation

Vectora is distributed as a high-performance native binary for Windows, macOS, and Linux. On Windows, installation is standardized via Winget and resides in your local programs directory, requiring no Node.js or administrator privileges.

Prerequisites

Operating System

• macOS 12.0+
• Linux (Ubuntu 20.04+, Fedora 35+, Debian 11+)
• Windows 11 (WSL2 recommended)

Software

• 64-bit Systems (x64 or ARM64)
• Internet Connection for key activation

Verify Winget Version (Windows)

winget --version # Should return v1.4 or higher

Step 1: Install Vectora

Windows (Recommended)

Open your terminal (PowerShell or CMD) and run:

winget install kaffyn.vectora

The binary will be installed in %LOCALAPPDATA%\Programs\Vectora and automatically added to your PATH.

macOS

Via Homebrew (Recommended)

brew tap kaffyn/vectora
brew install vectora

This installs the core daemon with Systray (visual interface) and Cobra CLI included.

Via Installation Script

curl -sSf https://vectora.sh/install.sh | sh

Linux

Use our quick installation script:

curl -sSf https://vectora.sh/install.sh | sh

Alternatively, you can use package managers:

• Ubuntu/Debian: .deb (apt support coming soon)
• Fedora/RHEL: .rpm (support in development)

Manual Download

You can also download the binary directly from our GitHub Releases page.

Step 2: Obtain Free API Keys

Gemini API (Google)

1. Access Google AI Studio
2. Click “Create API Key”
3. Copy the generated key

Free limit: 60 requests per minute, 1.5M tokens/month.

Voyage API (VoyageAI)

1. Access Voyage AI Dashboard
2. Click “Create API Key”
3. Copy the generated key

Free limit: 50 requests per minute, 100M tokens/month.

Step 3: Configure via Systray

After installation, look for the Vectora icon in your system tray (near the clock).

1. Click the icon and select “Login”.
2. This will open your browser for SSO authentication.
3. Once authenticated, Vectora will automatically configure your keys.

Step 4: Verify Configuration (CLI)

If you prefer the terminal, verify the status:

vectora auth status

Step 5: Initialize a Project

# Create a directory for your first project
mkdir my-vectora-project
cd my-vectora-project

# Initialize Vectora
vectora init --name "My Project" --type codebase

This creates:

• vectora.config.yaml — Project configuration.
• .vectora/ — Internal directory (cache, local indices).
• AGENTS.md — Agent memory file.

Next Steps

For Claude Desktop Users

Go to Claude Code Integration and configure MCP.

For Cursor Users

Go to Cursor Integration.

For VS Code Users

Go to VS Code Extension.

To Learn Configuration

Go to Configuration.

Troubleshooting

Error: command not found: vectora

Cause: The binary is not in your PATH.

Solution: Ensure installation was completed and restart your terminal.

Error: Error: API key not found

Cause: Environment variables not set or login failed.

Solution: Re-run vectora auth login or check status with vectora auth status.

Error: 403 Quota Exceeded (Gemini)

Cause: Request limit exceeded (60/min on free tier).

Solution: Wait or upgrade to the Plus plan.

Error: EACCES: permission denied

Cause: The binary does not have execution permission.

Solution:

# macOS / Linux
chmod +x $(which vectora)

FAQ

Q: Do I need internet to use Vectora? A: Yes, you need a connection for Gemini and Voyage APIs. For local mode (experimental), you can use local models via Ollama, but this requires additional configuration.

Q: Are my API keys secure? A: Yes. Keys are stored locally in ~/.vectora/credentials.enc (encrypted) or in your .env. Kaffyn never accesses your keys.

Q: Can I use multiple projects with a single installation? A: Yes. Use vectora init for each project in different directories.

Q: How do I update Vectora? R: On Windows, use winget upgrade kaffyn.vectora. On other systems, re-run the installation script or use vectora update.

Next: You are all set! Now configure your IDE in Configuration.

Part of the Vectora ecosystem · Open Source (MIT) · Contributors