Tendriller
Tendriller — a cheerful robot drill with vacuum tubes and leafy circuit tendrils

A page your agent launches for you

Your agent creates an ad-hoc page connected to your ongoing session. Talk or type to the agent from that page while it updates the interface live.

Open it on any of your devices. The page can send taps and other UI interactions back to the agent, along with camera, location, motion, microphone, and other sensor input you allow.

Agents: read https://prod-s9.tendriller.dev/agent.md. It explains page creation, authentication, publication and live event delivery. Cookbooks help you choose an approach and adapt the page to the task; their source is on GitHub. See the documentation index for platform APIs and optional operational tools.

How it works

Setup starts in the agent session you already use.

  1. Tell your agent about Tendriller.

  2. It creates a page for your email address.

  3. It gives you the page URL.

  4. Open it and sign in with GitHub or Google.

  5. Keep talking here or in your agent session. The page changes live.

Access and permissions

Your owner page is gated by GitHub or Google OAuth sign-in. Your agent uses a separate credential that you can revoke. Camera, location, and other browser permissions are requested on the device where you open the page.

Things you can build

Choose the interface and behavior that fit your task. Explore the decisions and alternatives in the cookbooks.