Model Context Protocol (MCP) Security Gateway
The open security gateway for AI coding agents. Enforces identity, deterministic OPA policies, and audit ledgers on every tool invocation.
1. What is the Model Context Protocol (MCP)?
The Model Context Protocol (MCP) is an open standard introduced by Anthropic that enables language models to securely interact with local files, development tools, databases, and enterprise APIs. However, raw MCP execution grants AI assistants unrestricted tool access. Without governance, an agent can perform unauthorized cloud modifications, delete repositories, or leak proprietary secrets.
2. The Kavach Security Gateway Layer
Kavach acts as a secure reverse-proxy and authorization firewall in front of your MCP servers. Every tool call must pass through eight distinct deterministic gates before execution:
3. Transport Protocols & Compatibility
The gateway natively supports both Server-Sent Events (SSE) streaming transport for remote cloud assistants and Standard I/O (stdio) transport for local developer IDEs like Cursor and VS Code.
