Bibliography

Reading Contract: Use this bibliography as evidence inventory. Separate public source anchors, product references, and supporting material before citing them in your own notes.

Primary Codex Sources

• OpenAI, Introducing Codex, May 16, 2025. Used for the public framing of Codex as a software-engineering agent with isolated task environments, file editing, command execution, evidence, and review workflows.
• OpenAI, Codex is now generally available, October 6, 2025. Used for the product evolution story across terminal, editor, cloud, Slack, SDK, and admin surfaces.
• OpenAI, Introducing upgrades to Codex, September 2025. Used for careful context on Codex-specific models and agentic software-engineering workflows.
• OpenAI, openai/codex. The public source repository analyzed by this book.
• OpenAI Developers, Docs MCP. Used as an OpenAI-maintained example of connecting Codex to MCP servers.

Agents and Tool Use

• Shunyu Yao et al., ReAct: Synergizing Reasoning and Acting in Language Models, 2022. Used as conceptual background for interleaving reasoning, actions, observations, and further decisions.
• Google Research, ReAct: Synergizing Reasoning and Acting in Language Models, November 8, 2022. Used as an accessible author-side explanation of ReAct.
• Timo Schick et al., Toolformer: Language Models Can Teach Themselves to Use Tools, 2023. Used as broader context for language models learning when and how to use tools.
• OpenAI, Function calling and other API updates, June 13, 2023. Used for historical background on structured tool/function calls in OpenAI APIs.

MCP and Integration

• Anthropic, Introducing the Model Context Protocol, November 25, 2024. Used for MCP’s public motivation.
• Model Context Protocol project, Specification. Used for protocol-level claims about MCP clients, servers, capabilities, and JSON-RPC.
• Model Context Protocol project, Architecture. Used for the host-client-server explanation.

Rust, TUI, and Async Background

• Ratatui project, Ratatui. Used for terminal UI background.
• docs.rs, ratatui crate documentation. Used for widgets, layout, frames, and terminal rendering concepts.
• Tokio project, Async in depth and Channels. Used for async and message-passing background.

Security and Publishing

• OWASP, Top 10 for Large Language Model Applications. Used for risk vocabulary around prompt injection, excessive agency, and tool-related threats.
• NIST, Artificial Intelligence Risk Management Framework 1.0, January 26, 2023. Used for general risk-management language.
• containers/bubblewrap, README. Used for Linux sandboxing context.
• GitHub Docs, About custom domains and GitHub Pages. Used for project-site custom-domain behavior under a user site.
• Claude Code from Source, Architecture, Patterns & Internals. Used only as a benchmark for book structure: part-based navigation, learning outcomes, process transparency, visual architecture maps, and chapter-level transferable patterns.