#104104th of 149 in Skills, plugins and rules
awesome-codex-subagents
Collection of 175+ ready-made Codex subagent definitions as TOML files
- Stars
- 6.3k
- License
- MIT
- Last commit
- Oct 2026
2,618 on 25 Mar 20266,328 on 14 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
A library of Codex custom agent files in .toml format, grouped into 13 categories such as core development, language specialists, infrastructure, quality and security, and data and AI. You copy the files you want into ~/.codex/agents/ or .codex/agents/. Each file sets a model, reasoning effort, sandbox mode and instruction text. Codex does not spawn them automatically, so you delegate in prompts.
Who it is for: Codex users who want prebuilt specialist subagents to drop into their projects
Strengths
- Plain .toml files installed by copying into ~/.codex/agents/ or .codex/agents/
- Each agent sets model, reasoning effort and sandbox mode (read-only or workspace-write)
- Covers many languages and frameworks plus infrastructure, security and data roles
- Project-level agents override global ones on name conflicts
Weaknesses
- Codex does not auto-spawn custom subagents; you must delegate explicitly in prompts
- No installer or package manager; installation is manual file copying
- README gives no benchmarks of how the agents affect output quality
- Model names in the files are fixed and may not match your Codex setup
What it needs
- no GPU
- Needs Codex
- Models: gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna
Also in Skills, plugins and rules
See all 149| Rank | Project | Score |
|---|---|---|
| 1 | cavemanCuts token use in coding agents with terse replies and compressed inputs | 95 out of 100 |
| 2 | skillsAgent skills for planning, TDD, debugging and code review | 94 out of 100 |
| 3 | eccAgents, skills, hooks and rules bundle for Claude Code, Codex and other harnesses | 94 out of 100 |
| #4 | impeccableDesign skill with 24 commands and a detector for AI-generated frontends | 94 out of 100 |
| #5 | ponytailSkill prompt that makes coding agents write less code and add tests | 92 out of 100 |