peta-core

Overview

Peta Core is a comprehensive control plane for MCP (Model Context Protocol) that serves as a gateway, vault, policy engine, and audit trail between AI agents and downstream MCP servers. It provides essential infrastructure for running MCP in production, including credential vault with encrypted storage, RBAC/ABAC policy engine with optional human-in-the-loop approvals, and detailed audit logging of every tool call. The solution supports managed runtime with lifecycle controls for downstream servers, protocol compatibility, and self-hosted deployment model without SaaS dependency.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• catalog.search

  Search the tool catalog for available tools across servers

  Try:Find tools for file operationsTry:Search for data analysis tools
• catalog.describe

  Get detailed description of a specific tool from the catalog

  Try:Describe the server.create_file toolTry:Get details of database query tools
• catalog.execute

  Execute a tool discovered through the catalog

  Try:Execute the git.create_issue tool with the described parametersTry:Run the data.analyze tool with specified options
• tools.list

  List all available tools for the current server

  Try:Show me all available toolsTry:List the tools I can use
• tools.call

  Execute a specific tool with provided arguments

  Try:Call the weather.get tool with location 'New York'Try:Execute the file.read tool with path '/data/report.txt'
• policy.evaluate

  Evaluate a policy decision for a tool call

  Try:Check if I can access the financial databaseTry:Evaluate if I can delete files
• approval.request

  Request approval for a high-risk operation

  Try:Request approval to delete customer recordsTry:Get approval for system configuration changes
• approval.status

  Check the status of a pending approval request

  Try:What's the status of my approval request?Try:Check if my deletion request was approved

Note: Tool names inferred from the architecture documentation describing MCP gateway functionality and catalog features. Exact tool signatures not explicitly documented in this README.

Comparable tools

Installation

docker run -d --name peta-core -p 8080:8080 \
  -e DATABASE_URL=postgresql://user:pass@localhost:5432/peta \
  -e NEXTAUTH_SECRET=your-secret-key \
  petaio/peta-core:latest

npm install -g @peta/core
peta-core start

{
  "mcpServers": {
    "peta": {
      "command": "npx",
      "args": ["@modelcontextprotocol/server-peta", "--host", "localhost", "--port", "8080"]
    }
  }
}

FAQ

What databases are supported?

Peta Core requires PostgreSQL 15+ as its primary database. This is for storing audit logs, policy configurations, and credential vault information.

How does Peta Core handle credentials?

Credentials are encrypted at rest using PBKDF2 + AES-GCM and are injected server-side at execution time. Clients never see raw credentials, and secrets are never included in audit logs.

Compare peta-core with