codex-cli-best-practice

Name: codex-cli-best-practice
Rating: 77 (1 reviews)
Author: shanraisshan

shanraisshan/codex-cli-best-practice

A comprehensive best-practices reference repository demonstrating advanced agent orchestration, configuration management, and skill development for Codex CLI. It provides working examples of TOML profiles, subagents, MCP servers, hooks, and reusable instruction packages.

965 64MITUpdated 2026-06-04

Great

View on GitHub Open source file

Overview

This repository serves as a blueprint for agentic engineering with Codex CLI, illustrating the Agent-to-Skill orchestration pattern. It includes structured guides, configuration profiles, custom agents, and a fully functional weather data system example that fetches live data and generates SVG cards using progressive skill disclosure.

Capabilities

▸Layered configuration profiles (conservative, development, trusted, ci, review)
▸Subagent registration and parallel execution orchestration
▸Reusable skill packaging with progressive disclosure and YAML frontmatter
▸MCP server integration with OAuth and parallel tool calling support
▸Event-driven hook integration for logging and automation

Best for

Configuring layered TOML settings and security profiles for Codex CLI, Orchestrating subagents and reusable instruction skills for complex workflows, Integrating Model Context Protocol (MCP) servers and hooks into development pipelines, Learning best practices for agentic codebase interaction and project documentation

Works with

Codex

Evoa Score breakdown= Σ (score × weight)

Task usefulness20%88 → +17.6

実タスクにどれだけ役立つか（機能の豊富さ・用途の明確さ）。 — AIによるcapabilities/use-cases解析

Code quality15%90 → +13.5

実装・指示の品質。 — AIによるSKILL.md/README解析

Maintenance15%85 → +12.8

リポジトリがどれだけ活発に保守されているか。 — GitHub 最終push日時の新しさ

Documentation12%44 → +5.3

ドキュメントの充実度・分かりやすさ。 — README/独自要約の情報量

Security15%80 → +12.0

危険・不審な挙動が無いか。 — AIによるセキュリティレビュー

Originality10%85 → +8.5

ありふれたラッパーではない独自性。 — AIによる独自性判定

Popularity8%76 → +6.1

コミュニティの採用度。 — GitHub Stars/Forks（対数スケール）

Compatibility5%25 → +1.3

対応AIエージェントの広さ。 — AIによる対応エージェント判定

Weighted total77.0 / 100

ライセンス不明/制限あり（Red）のSkillは総合スコアに0.85倍の補正を適用します。ランキングはこのScoreのみで決まり、広告で変わりません。算出方法の詳細 →

Security considerations

The repository configures sandbox environments and approval policies through TOML files. Users should ensure appropriate security profiles (such as read-only or on-request approval) are used when executing untrusted code.