fcakyon/claude-codex-settings
Battle-tested Claude Code, OpenAI Codex, Cursor configs, plugins, hooks and agents with Kimi MiniMax and GLM API support.
What it solves
This project provides a collection of battle-tested configurations, plugins, and guidelines to improve the performance and output quality of AI coding tools like Claude Code, Cursor, Codex CLI, and Gemini CLI. It specifically targets common AI pitfalls such as over-engineering, bloated abstractions, and the use of generic AI buzzwords in generated text.
How it works
It operates as a set of plugins and hooks that can be installed into various AI CLI tools. These plugins add specific "skills" (capabilities), commands, and automation hooks that intercept or modify the AI's behavior. For example, it uses Python scripts as hooks to block AI buzzwords before text is saved or provides specialized agents for second opinions on architectural decisions.
Who it’s for
Developers using AI-powered coding assistants (specifically Claude Code, Cursor, Codex CLI, and Gemini CLI) who want more professional, concise, and human-like code and documentation.
Highlights
- Simplify Plugin: Reviews changed code for redundancy and over-engineering using parallel agents.
- Humanize Plugin: Blocks a curated list of 53+ AI buzzwords and cliches using a hook to ensure natural writing.
- Advisor Plugins: Integrates second-opinion agents (Fable 5 and GPT) to challenge plans before commitment.
- ADHD Output Style: Reduces token usage and filler by enforcing a concise, answer-first format.
- Intelligent Compact: Prevents the AI from losing critical context (like file paths and root causes) during session summarization.
- Office Skills: Adds capabilities for processing PDF, Word, PowerPoint, and Excel files.