llms.txt Content
# Metorial
## What Metorial is
**Metorial** is the **integration platform for agentic AI**. Modern applications need models to **read and write** real systems: calendars, CRMs, issue trackers, databases, messaging, internal APIs, and thousands of SaaS products. Hand-rolling HTTP clients, OAuth flows, token refresh, schema drift, and retries for every provider does not scale. Metorial gives you a **single, productized layer** between your agents and that long tail of tools—so your team ships features instead of maintaining a private integration zoo.
The platform is built on the **Model Context Protocol (MCP)**, an open standard for how models discover and invoke tools. Metorial runs **MCP servers and sessions** as first-class infrastructure: you get consistent semantics for listing tools, calling them, passing context, and handling failures—whether you are in a hackathon prototype or running **large numbers of concurrent MCP connections** in production.
**Authentication and governance** are central, not bolted on. Metorial supports **OAuth**, **API keys**, and patterns for **end-user** and **service** credentials so secrets are not scattered across your codebase. You can model **projects**, **organizations**, **environments**, and **deployments** so the same integrations behave correctly across staging and production.
**Operations** are built for builders: a **hosted dashboard** (and full **HTTP API**) for API keys, **server deployments**, live **sessions**, and **monitoring**—including **session recording** and **error reporting** when a tool call fails—so you can debug what the model actually did, not guess from logs alone. An **embedded MCP Explorer** lets you exercise servers before you commit them to your app.
Metorial is **open source at the core** ([github.com/metorial/metorial](https://github.com/metorial/metorial)) with a **managed cloud** option at [app.metorial.com](https://app.metorial.com/) when you want the fastest path to reliability and