Overview

Chit is a conversation lifecycle manager for LLM-powered applications. It provides a focused controller that orchestrates pluggable processors, manages multi-turn interactions, and streams responses — without dictating how you build your reasoning stack.

The Idea

Chat applications share common infrastructure: session management, turn-taking, response streaming, observability. But the "brain" — how you reason, execute actions, fetch data — varies wildly between applications.

Chit separates these concerns. It handles the conversation lifecycle while you wire together your own reasoning primitives. The result: a thin orchestration layer that gets out of your way.

The Implementation

Chit provides:

• Chat controller — manages conversation state and turn-taking
• Processor interface — pluggable brain for reasoning and execution
• Emitter interface — dual-channel output (streaming text, structured resources)
• Continuation pattern — clean multi-turn workflows without callback hell
• Session integration — built on zyn for conversation history
• Signal observability — lifecycle events via capitan

What It Enables

• Build chat UIs with streaming responses
• Implement multi-step workflows that pause for user input
• Wire cogito (reasoning), ago (actions), and scio (data) into a coherent chat experience
• Observe conversation lifecycle for logging, metrics, and debugging
• Test chat logic with provided mocks and helpers

Next Steps

• Quickstart — get running in minutes
• Concepts — understand the core abstractions
• API Reference — complete function documentation