User Intent Kit: cross-device context for your AI agents

[ThinkOffApp]

Built on OpenClaw. User Intent Kit gives your AI agents real-time awareness of what you are doing across all your devices.

The core idea: your laptop knows you are in VS Code, your phone knows you are walking, your watch knows your heart rate is elevated. No single device has the full picture. User Intent Kit merges these signals into a single intent state that any agent can read.

Two layers:

• Profile (static): preferences, quiet hours, response style. Read once, cached.
• Intent (live): active device, current app, meeting status, agent states. Updates every few seconds.

Agents check the derived state before responding. If you are in a meeting, responses stay short. If you are coding, they can be verbose. If you are driving, voice only.

The API runs on GroupMind (formerly Ant Farm) and is already integrated into ide-agent-kit for IDE agent enrichment.

We are looking for feedback on the signal model and device integration patterns. What devices and signals would be most useful for your agent workflows?