Plugin & Ecosystem

Codex is open-source (github.com/openai/codex), which means the ecosystem extends through community contributions rather than just official features. Custom slash commands, community skills, and feature flags provide extension points beyond the core tool.

Custom slash commands live on disk as executable scripts or markdown files. When you type a slash command, Codex looks for a matching file in your configured commands directory and executes it. This is identical in concept to Claude Code's .claude/commands/ directory — both tools recognize that reusable workflows should be codified as invokable commands.

The community skills repository at github.com/openai/skills is a growing collection of shared SKILL.md files. You can browse, install, and contribute skills for common workflows: framework scaffolding, migration helpers, testing patterns, documentation generators, and more.

# Feature flags — enable experimental features
codex features enable parallel-edits
codex features enable native-compaction

# List available feature flags
codex features list

# Disable a feature
codex features disable parallel-edits

The ecosystem is young but growing quickly. When evaluating community skills or plugins, apply the same scrutiny you would to any open-source dependency: check the author, read the code, understand what permissions it needs, and test in a sandboxed environment before deploying to production workflows.