Skip to main content

Autonomous Inbox — API, MCP & CLI Documentation

Welcome! Autonomous Inbox is Kobana's service for receiving, storing, filtering and processing inbound email with AI agents. This site documents three ways to reach it: the public REST API under /api/v1 — mTLS + Bearer JWT, JSON, cursor pagination and webhooks — the remote MCP server, for AI clients that would rather call tools than HTTP endpoints, and the kobana CLI, for calling it straight from a terminal or a script.

Use the sidebar to navigate, starting with Introduction — or go straight to the API Reference if you already know what you're looking for. Connecting an AI assistant instead? See the MCP server. Prefer the terminal? See the kobana CLI.

Getting started​

  1. Get authenticated — mTLS certificate + Bearer JWT.
  2. Make your first call against /api/v1/workspaces.
  3. Create an Inbox and start receiving email.
  4. Subscribe a Webhook to react to events in real time.

What you'll find here​

  • Overview — authentication, endpoints, specifications, rate limiting, pagination, filtering, idempotency, errors and data retention.
  • Models — Workspace, Inbox, Email, Agent, AgentRun, Webhook, WebhookDelivery and SystemEvent, the domain objects used across the API.
  • Workflows — how an agent can end its run with a chain of HTTP calls against your systems: how it works, a quickstart, the step and placeholder reference, and troubleshooting.
  • Resources — the full endpoint reference for every resource under /api/v1, generated from the OpenAPI specification, with request/response schemas and a try-it-out console.
  • MCP — the remote Model Context Protocol server, which exposes the same operations as tools an AI client can call: authentication, transport, and every tool with its parameters.
  • CLI — kobana, the single command-line tool for every Kobana product's API (this one included): installation, authentication, usage and schema introspection.