Skip to main content

Getting Started

Get Peta Running in 5 Steps

Connect AI agents to MCP tools through a secure gateway — with built-in secret management and approval workflows.

STEP 1

Install & Deploy

Run a single script — it validates your environment, configures ports, and launches all services.

1.1

Download deployment script

Download the script for your operating system:

Terminal
$curl -O https://files.peta.io/installation/deploy-peta-mac.sh
$chmod +x deploy-peta-mac.sh
1.2

Run the script

Execute and follow the interactive prompts:

Terminal
$./deploy-peta-mac.sh
1.3

Choose deployment mode

The script prompts for deployment mode: Core + Console (recommended), Core Only, or Console Only.

Deployment complete

On success, you'll see access info:

Core API: http://localhost:3002
Core Health: http://localhost:3002/health
Console Web: http://localhost:3000

STEP 2

Configure the Console

2.1

Open Console

Open your web browser and navigate to http://localhost:3000

2.2

Initialize Vault

Set up your master password through the console setup wizard. This password initializes your vault and secures Owner Token generation.

Initialize Vault Setup
2.3

Backup Owner Token (Critical)

Owner Token appears once after initialization. Copy it immediately and store in a secure password manager.

Backup Owner Token
2.4

Create Access Token

Generate an Access Token in Console → Access Token Management → Create Token. Keep Owner Token for admin tasks only and use Access Token for daily operations.

Create Access Token

STEP 3

Add MCP Tools

Choose how to add MCP tools to your gateway.

OPTION A

Deploy Managed Templates

Pre-built tools (GitHub, Slack, PostgreSQL, etc.) ready to deploy in one click.

Console path: Tool Configuration to Add Tool to Templates

OPTION B

Connect Existing MCP Servers

Point Peta at your existing MCP server URLs and apply access policies.

Console path: Tool Configuration to Add Tool to MCP Server

OPTION C

Wrap REST APIs (No-Code)

Turn any REST API into an MCP tool with a visual mapping interface.

Console path: Tool Configuration to Add Tool to REST API

STEP 4

Configure MCP Client

Add this to your MCP client config (Claude Desktop, Cursor, or any MCP-compatible tool).

Add to your MCP client settings:

mcp_config.json
{
"mcpServers": {
"peta": {
"url": "https://your-host:3002/mcp",
"headers": {
"Authorization": "Bearer <ACCESS_TOKEN>"
}
}
}
}

url — Your Peta Core endpoint. Use http://localhost:3002/mcp for local setup.

Authorization — The Access Token created in Step 2.4. Use HTTPS host only for remote access.

VERIFY

Check Your Setup

Confirm everything is running correctly.

Check Core health:

Terminal
$curl http://localhost:3002/health

Expected response:

Response
{ "status": "ok" }

Open Console in your browser:

URL
http://localhost:3000

All set!

If both checks pass, your Peta infrastructure is ready. Open Console at http://localhost:3000 to start managing tools and tokens.

STEP 5

Install Peta Desk (Optional)

For teams that need a desktop GUI to manage MCP connections — or a human-in-the-loop controller for approving sensitive operations. AI agents connect directly via API; skip this step for API-only setups.

macOS (v1.0.1. 7 Mar 2026)
Windows (v1.0.1. 7 Mar 2026)
Linux — Coming Soon

When to use: Install for team members who need a visual interface to manage MCP connections, or when you need human approval gates for sensitive operations. AI agents connect via API without Peta Desk.

Human-in-the-Loop: Mark specific tool actions as requiring approval — Peta Desk delivers real-time notifications when those actions are triggered by AI agents, letting a human review and approve or reject before execution.

Peta Desk Human Approval Interface

TROUBLESHOOTING

Common Issues

Re-run the deployment script and choose available ports when prompted. Verify assigned ports with docker compose ps in your deployment directory.

NEXT STEPS

What's Next?

Explore Architecture

Understand how Peta's security model protects your infrastructure

View Architecture →

See Use Cases

Discover how teams across industries deploy Peta MCP

Browse Use Cases →

Get Support

Need help? Join our community or request early access

Get Help →