The missing piece for agents
An AI assistant can write a script that checks a page, summarizes a feed or calls an API. When the chat ends, the work stops. ClickRuntime is the runtime underneath: the assistant saves the job, turns it on with your approval, and we keep it running on a schedule or whenever an outside event arrives.
Approval is built in
An agent can build, turn on, run, test and fix an automation until a run succeeds. Each of those actions goes through your AI assistant's own permission prompt, so you approve every step. Before running something live, the agent tells you what it will do in outside services. Keys and secrets are added only by you in the ClickRuntime dashboard, and deleting an automation needs a separate confirmation.
Keys never pass through the model. When a job needs one, the assistant sends you to the ClickRuntime dashboard to save it, and the script reads it from the environment at run time.
Works with any MCP assistant
ClickRuntime is an MCP connector. Add it once to ChatGPT, Claude, Claude Code, Cursor or any assistant that supports remote MCP servers with OAuth, sign in, and ask in any conversation. The server is https://clickruntime.com/api/mcp. The connect guide shows the steps for each assistant.
What we refuse to run
The assistant declines jobs that are harmful or that we can't run responsibly: malware, secret monitoring of a person, collecting credentials, getting around another service's access controls or rate limits, moving money or placing trades, and collecting health, government ID or payment card data.