awesome-codex-subagents and awesome-skills
These are ecosystem siblings—one is a curated directory/index of agent skills across multiple platforms, while the other is a specialized implementation providing 130+ ready-made Codex subagents that would be cataloged or referenced by such a directory.
About awesome-codex-subagents
VoltAgent/awesome-codex-subagents
A collection of 130+ specialized Codex subagents covering a wide range of development use cases.
Each subagent is defined in Codex-native `.toml` configuration files with smart model routing (gpt-5.4 for deep reasoning tasks, gpt-5.3-codex-spark for lighter work) and sandbox mode controls that match the agent's role—read-only for auditors and reviewers, workspace-write for developers. Organized across 10 categories spanning core development, language specialists, frameworks, and infrastructure, agents install into `~/.codex/agents/` (global) or `.codex/agents/` (project-scoped with override precedence) and must be explicitly delegated in prompts.
About awesome-skills
gmh5225/awesome-skills
A curated list of Agent Skills, resources, and tools for AI coding agents like Claude Code, Codex, Gemini CLI, GitHub Copilot, and more.
The collection organizes Skills using a **progressive disclosure architecture** where agents first scan lightweight metadata (~100 tokens) to identify relevance, then load full instructions (<5k tokens) and bundled resources on-demand—minimizing token overhead. Skills are standardized as folders with `SKILL.md` metadata files plus optional scripts, templates, and resources, enabling dynamic discovery across 15+ agent platforms (Claude Code, Copilot, Cursor, Gemini CLI, Codex, and conversational tools like Coze). The registry maps platform-specific skill directories and includes official implementations from Anthropic, OpenAI, HuggingFace, and Vercel teams.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work