MCP & AI Agents

Connect an AI agent to Sonta over MCP — the same control surface your team uses, scoped by role.

• GuideWorking with Sonta programmaticallySonta is built to be driven programmatically, not just clicked. An AI agent or your own code connects over MCP and gets the same control surface your team has — building collections and fields, reading and writing entries, configuring automations and booking — scoped by the role it connects as. Purpose-built tools cover the common work; the rest of the surface is available for the long tail.
• ArticleHow an agent connects and what it can reachAn agent connects by being added to your AI assistant as a connector; you sign in to Sonta once, through the assistant, and the assistant holds that authorization from then on. It works in one business at a time, remembered on your profile between sessions, and it can do exactly what your role permits. Nothing more.
• ArticleWhat the tools coverThe connection offers purpose-built tools for the work that comes up constantly — reading a collection's shape, finding entries, writing them with validation, checking booking readiness, looking up automation options, inspecting a run, and installing a blueprint — plus step-by-step recipes for multi-step tasks. Prefer them over improvising: they validate input and explain failures.
• ArticleConnect to the MCP serverTo connect an AI assistant to Sonta, add Sonta's MCP server to the assistant as a connector and sign in to Sonta when the assistant asks. From then on the assistant acts as you, with your role's permissions, in the business you have selected.
• ArticleRead shape vs write shapeMany fields are written in one shape and read back in another. You write a reference as the target's id and read back the whole entry; you write an option as its key, slug, or label and read back its key with the label and slug. Send what the field expects on write, and never send a read value back without converting it.