Klavis AI

Open-source MCP infrastructure — 50+ hosted MCP servers with built-in OAuth so AI agents can connect to GitHub, Slack, Notion, Gmail, and more without writing auth code

FreemiumOpen Source Agents & Automation

Free Hobby plan; Pro/Team plans from $99/month; Enterprise available

Visit Tool

Overview

Klavis AI is a managed MCP infrastructure platform that provides 50+ hosted MCP servers with built-in OAuth, eliminating the complexity of self-hosting integrations for AI applications. It's designed for developers building LLM-powered products that need to connect to real-world tools and data sources.

Key Features

50+ pre-built MCP servers covering popular tools and APIs
Managed OAuth handling — no custom auth logic required
Supports Slack, Google Drive, GitHub, Notion, and many more
Simple API to provision and configure integrations per user
Scales automatically without infrastructure management
Compatible with any MCP-enabled AI framework or client

Pricing: Free tier available; usage-based pricing for production deployments.

Pros

50+ production-ready MCP servers covering the most common business tools
Auth built in — OAuth and user-based access controls managed for you
Open source and self-hostable via Docker for full data control
MCP-compatible — works with Claude, Cursor, and any MCP-enabled agent framework
Tools can be added or removed at runtime without redeployment

Cons

Pro plans start at $99/month — can be steep for indie developers
Relatively new (YC X25) — integrations and documentation still maturing
Dependent on Klavis infrastructure unless you self-host

Product Updates

Klavis AI (YC X25)@Klavis_AI

Introducing Klavis MCP Sandbox-as-a-service ! RL environments are becoming the new labeled datasets. But here's the dirty secret: training models to use real SaaS tools (Gmail, Salesforce, Slack, Jira) is an infrastructure nightmare. You need: Hundreds of authenticated test