MCP Server Overview
MCTL exposes a Model Context Protocol (MCP) server that lets AI assistants manage your infrastructure through natural language.
What is MCP?
MCP is an open protocol that connects AI models to external tools and data sources. Instead of copy-pasting commands, you describe what you want in natural language and your AI assistant calls the appropriate tools.
Server Details
| Endpoint | https://api.mctl.ai/mcp |
| Transport | Streamable HTTP (POST/GET) |
| Auth | OAuth 2.0 PKCE or Bearer token |
| Tools | 39 (21 read-only, 14 write, 4 destructive) |
Capabilities
With the MCTL MCP server you can:
- Manage tenants — create, list, inspect, and delete tenants
- Deploy services — deploy, scale, rollback, and retire services
- Monitor operations — track async operations and workflows
- Handle incidents — list, inspect, acknowledge, and resolve incidents
- Manage domains — add, verify, and remove custom domains
- Provision databases — create databases with auto-configured secrets
- Preview environments — create and manage preview deployments
- Resource management — check usage, get sizing recommendations
What You Get
Once connected, your AI assistant can:
- Inspect platform state — query services, tenants, resource usage, logs, and workflow status in real time
- Deploy and manage apps — onboard, deploy, rollback, and preview services; create workspaces; provision databases
- Manage domains and scaling — add custom domains with auto-TLS, configure autoscaling, verify DNS
- Full audit trail — every write operation creates a git commit in the GitOps repo for complete traceability
How Access Works
Authentication follows a three-step model:
- Token validated per request — your GitHub token or OAuth credential is checked on every MCP call. The server never stores tokens.
- Team-scoped access — access is limited to workspaces resolved from your GitHub team memberships in the GitOps repo. Admins can operate on all tenants.
- Full audit trail — every write operation submits an Argo Workflow that produces a git commit. All changes are traceable and reversible.
For detailed auth information, see Authentication.
Tool Annotations
Every tool is annotated with its behavior:
| Annotation | Meaning |
|---|---|
readOnly | Only reads data, no side effects |
destructive | Modifies or deletes resources |
See the full Tools Reference for details on each tool.