An AI agent connected to Thinger.io can work with your account the way you do in the console: it reads device data, creates and configures resources, and builds dashboards. It connects through the Model Context Protocol (MCP), the standard way for AI clients such as Claude Code, Claude Desktop or Cursor to use external tools.

Today you connect through the CLI

The MCP server runs with the thinger CLI: you sign in once with thinger login, and your AI client starts thinger mcp. See Connect an AI agent.

How it works

The MCP server gives the agent a set of tools, such as thinger_resource_list or thinger_bucket_data_read. Each time the agent uses a tool, the server calls the REST API of your Thinger.io server with the access token of your CLI profile. The agent can do exactly what that token allows, nothing more: see Permissions.

Most clients ask you to approve a tool call before it runs. Review the calls that change or delete data before you approve them.

What you can ask an agent

Operate devices. List devices and their state, read and call device resources, change properties, move devices between products.

Ask your AI agent

Which of my devices are disconnected right now, and when did each one last connect?

Analyze data. Query Data Buckets with aggregation, compare devices and explain trends.

Ask your AI agent

Show me the hourly average temperature that YOUR_DEVICE stored in its Data Bucket over the last 24 hours, and tell me when it went above 30 °C.

Configure the platform. Create and change products, endpoints, alarm rules, access tokens, asset types and groups, projects and file storages.

Ask your AI agent

Create an alarm rule that raises a warning when the battery property of any device of the product greenhouse drops below 20, and notifies me by email.

Build dashboards. Create a dashboard app on the Dashboard SDK, develop it against real data and publish it. See Build dashboards with AI.

Ask your AI agent

Build a dashboard for YOUR_DEVICE with a chart of its temperature and humidity over the last week, and publish it as the dashboard of the device.

Administer a server. On a private instance, with an administrator token: user accounts, cluster hosts, custom domains, brands, OAuth2 clients, identity providers and proxies.

Ask your AI agent

I have pointed the DNS of iot.example.com to the server. Add it as a custom domain and check that it resolves.

Use plugins. Install, start and stop plugins, and call the API they serve.

Explore the section

Connect an AI agentInstall the thinger CLI, sign in, and connect Claude Code, Claude Desktop or any other MCP client to your Thinger.io account.PermissionsWhat an AI agent is allowed to do in your Thinger.io account, how to limit it with an access token, and how to use it safely.MCP toolsEvery tool the Thinger.io MCP server offers to AI agents, with its parameters, and every resource type the agents can reach.Build dashboards with AIAsk an AI agent to create a dashboard app on the Dashboard SDK, develop it against your real data and publish it.Docs for agentsThe Thinger.io documentation in plain markdown for AI agents (llms.txt, llms-full.txt and a .md file per page), and the guides the MCP server serves.