| name | codex-plugin-discovery |
| description | Use when a user asks what Codex plugins are available or can be used, wants plugin recommendations for a task, asks to find/search/discover installable plugins, asks about the plugin marketplace, or asks whether a plugin can help with a task. |
Codex Plugin Discovery
Discover candidate Codex plugins from the official openai/plugins and openai/role-based-plugins GitHub repositories.
Core Rule
Use only metadata indexed from:
https://github.com/openai/plugins
https://github.com/openai/role-based-plugins
This boundary is intentional, even under deadline or coverage pressure. Do not read Codex .tmp marketplace caches, installed plugin caches, or active-session tool caches. Those sources are out of scope.
Do not auto-install plugins. Recommend candidates and explain the evidence only. Do not claim this covers the full Codex App Plugin Directory.
Workflow
- If the user asks what plugins are currently enabled/installed in this session, answer from the current session's available plugin list first. Keep it concise.
- If the user asks broadly what plugins are available, can be used, or exist in the marketplace, do not list every indexed plugin. State that this skill can search
openai/plugins and openai/role-based-plugins, explain the coverage limit, and ask for the task or category they care about.
- If the user asks for recent, recently added, new, last-week, past-N-days, or 最近几天 plugins, route to
python3 scripts/list_recent_plugins.py. If the user provides a day count, pass --days N; otherwise use the default 7-day window. If index/plugins-index.json is missing or stale, run python3 scripts/build_index.py first. For mixed requests that ask for both recency and task recommendation, do not combine recency filtering with task recommendation in this first version; answer recent additions first and offer a separate task search.
