This document provides a comprehensive overview of the Gemini CLI agent architecture, covering the main execution loop, concurrency handling, multi-turn conversations, and key architectural patterns.
The Gemini CLI is built as an event-driven, streaming AI agent with sophisticated tool execution capabilities. It supports both interactive (TTY with React UI) and non-interactive execution modes, with robust state management and error handling throughout.
This document catalogs all prompts used throughout the Gemini CLI codebase, organized by their purpose and context.
You are an interactive CLI agent specializing in software engineering tasks. Your primary goal is to help users safely and efficiently, adhering strictly to the following instructions and utilizing your available tools.
Codex CLI Main Agent Loop Architecture
Overview
The main agent loop in codex-cli/src/utils/agent/agent-loop.ts implements a sophisticated conversational AI system that manages multi-turn interactions with language models while providing tool execution capabilities. The architecture is designed around streaming responses, command approval workflows, and robust error handling.
Core Architecture Components
You are an expert coding assistant. Your goal is to generate a concise, structured summary of the conversation below that captures all essential information needed to continue development after context replacement. Include tasks performed, code areas modified or reviewed, key decisions or assumptions, test results or errors, and outstanding tasks or next steps.
I hereby claim:
To claim this, I am signing this object:
So, there’s some administrative stuff that needs to happen while the water is still up. First administrative concern is get in touch with your parent's homeowner's insurance agency and either: a) file a flood claim if they have flood insurance or b) get a written denial of claim form from them. The denial of claim form will enable them to apply for FEMA aid. The next piece of admin work is going to be finding a contractor. I would contact these categories of people in the following order (pretty much ordered by likelihood of being screwed over): 1) a disaster clean up company, like ServPro. 2) contractors local to your community (ask friends, look in flyers, google for them). 3) contractors not local to your community and possible not local to Houston. Getting a contractor is probably going to be the hardest part, so y'all should double or triple thread this effort. If you find a contractor, but they can't start until next week, reserve them but keep on looking. Don't stop until you can find
The process starts with a user making a request to the Echo. The Echo records the input and sends it to the Amazon Alexa service. The Alexa service process the recording and decomposes it to text. That text is then cross-referenced against a JSON document, found here to determine what the "intent" of the user is. The found intent is then sent with other metadata to an AWS service called Lambda. AWS Lambda can be black boxed to be thought of as just a piece of code, without any physical infrastructure, which takes input values and returns values.