Observed arrival · 2026-08-26
Agent Apps: A Product Layer for AI Agents
An open protocol and runtime for giving AI agents installable applications with tools, interfaces, state, authentication, and inter-app actions.
Field notes
The project models an agent application as more than a callable function: its visible protocol description includes typed tools, permissions, structured results, interfaces, storage, authentication, health checks, and intents for app-to-app calls. Its local Station example listens on port 4100 and reports 79 installed apps with 10 running, while Gateway exposes them to MCP clients. The page also lists separate components for scaffolding, testing, deployment, and runtime hosting.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue