# World2Agent (W2A)

An open protocol that standardizes how AI agents perceive real-world data, so that installing a sensor gives an agent structured, real-time signals in a common schema.

World2Agent (W2A) is an open protocol, not a product, that standardizes how AI agents perceive the world: a sensor watches some data source and emits structured signals following the W2A schema, and an agent receives those signals and decides what to do, so sensors can be swapped freely because they all speak the same format. The project ships the first sensors itself, such as one for Hacker News stories and one for frontier AI lab posts, and expects further sensors to come from the community.

It ships native plugins for three agent runtimes: a Claude Code plugin installed from a plugin marketplace, a bridge package for Hermes that installs skills through its skill system, and a bridge for OpenClaw, each letting a user add a sensor with a short command and have its signals trigger a fresh agent run against a generated handler skill.

It is written in TypeScript and released under the Apache-2.0 license.

Website: https://world2agent.ai