untether-architecture

The Untether Telegram bridge processes user messages and routes them to an agent runner, which executes tasks and reports progress back. Events flow through the core system using the UntetherEvent abstraction to track status and send approval notifications.

Untether is configured using TOML files and environment variables to define system settings and engine backends. This configuration system manages how different AI coding agents are registered and integrated into the bridge infrastructure.

You register a new AI coding agent backend by integrating it with Untether's engine system. The architecture uses a Runner abstraction to execute agent tasks and report progress tracking events back to the Telegram interface.

The UntetherEvent model is a core abstraction used to represent events within the Untether architecture. It facilitates progress tracking and approval notifications by defining how data flows between the Telegram bot and the agent runner.

Yes, Untether tracks agent task progress and sends approval notifications through its event model. The Runner abstraction executes tasks and updates the Telegram bridge with status events, allowing users to monitor and approve actions.

Modifying the Untether core infrastructure requires understanding its event model, Runner abstraction, and engine backend registration. Developers must work within the constraints of its TOML configuration and environment variable system when altering data flow.