The week after our public launch, we’re shipping the thing you asked for most: a hosted MCP server. Connect it to your AI assistant and it can answer questions about your apps, builds, and add-ons directly from live platform state — no copy-pasting build logs into a chat window.
https://mcp.watasu.io/mcpWhat MCP is, in one paragraph
MCP (Model Context Protocol) is an open standard that lets AI assistants talk to external systems. Instead of you describing your infrastructure to the model, the model reads it. Claude, and a growing list of other clients, can connect to any MCP server — and as of today, Watasu is one of them.
Read-only, on purpose
The Watasu MCP surface can see but not touch. An assistant connected through MCP can read app status, build history, and add-on inventory. It cannot deploy, scale, change config, destroy anything — and it never sees config var values, secrets, database credentials, or billing data. Those are redacted or absent entirely.
This is a deliberate design choice. Reading platform state is something you can safely hand to any assistant, in any context, without thinking hard about what could go wrong. Acting on the platform isn’t — so if you want an assistant that deploys and scales, give it the Watasu CLI in an environment you control. MCP is the safe default for visibility.
Connecting takes a minute
Most MCP clients only need the URL. In Claude Code, it’s one command:
claude mcp add --transport http watasu https://mcp.watasu.io/mcpOr in any client that takes a JSON server config:
{
"mcpServers": {
"watasu": {
"type": "http",
"url": "https://mcp.watasu.io/mcp"
}
}
}The first connection opens a Watasu consent page in your browser, where you choose what the connection can see: all your teams, or specific teams only. Tokens are short-lived and refresh automatically, and you can revoke any connection from the dashboard. For scripted or headless clients, a Watasu API key works directly as a bearer token and inherits your access — apps you can’t see, the assistant can’t see either.
What you can ask
Once connected, the assistant reads resources like watasu://overview, watasu://apps/{app}/builds/{number}/logs, and watasu://apps/{app}/addons. In practice that turns into questions like:
- “Why did the last deploy of
billing-apifail?” — it reads the build log and points at the broken migration, instead of you scrolling through output. - “Give me a morning summary of the
acmeteam.” — every app, its live status, the latest build, and anything currently unhealthy. - “Did my push actually go live?” — coding agents can verify their own deploys before moving on to the next task.
- “Is it us or the deploy?” — an on-call assistant that checks the release status before anyone starts digging.
Everything respects your existing access. Team-scoped connections see only their teams, and resources you can’t reach return nothing — the assistant has your permissions, never more.
Get started
The MCP guide has the full resource list and client setup details. Connect it, ask your assistant how your apps are doing, and tell us what you’d like it to see next: info@watasu.io.