Codex/Plugin/Q/Can you make plugin reference material in both Codex plugins and Claude plugins?
Properties
- logseq-entity
- Logseq/Entity/Question
- see-also
- Agent/Skills
Can you make plugin reference material in both Codex/Plugins and Claude/Code/Plugins?
AI Answer
- Short answer: Yes. Both can distribute reference documents alongside a skill. A small
SKILL.mddescribes when the material is relevant and which files to read; the detailed material lives in supporting files. - Answer/Official from OpenAI: Build skills — Add supporting resources and Claude Code: Add supporting files.
Shared reference layout
- The same skill and reference files can be packaged for both hosts. For example:
- In
SKILL.md, name the trigger and link to each document, such as: “When answering questions about product terminology, read terminology.md.” The example link resolves inside the packaged skill directory. - OpenAI's portable package uses root
plugin.json;.codex-plugin/plugin.jsonremains a supported compatibility format. Claude Code uses.claude-plugin/plugin.jsonand discovers skills under rootskills/. The reference contents can stay shared while the manifests describe the package for each host. OpenAI package structure, Claude Code standard layout.
When the material is read
- Supporting files are read when needed. Bundling documents does not load all their contents into every conversation. Link the documents from
SKILL.mdand explain when to consult each one. - A reference skill can answer questions from documentation without running scripts or connecting to an MCP server. OpenAI explicitly supports skills containing only packaged instructions and resources; Claude Code describes reference material as a skill content type. OpenAI skills, Claude Code skill content types.
- Supporting files are read when needed. Bundling documents does not load all their contents into every conversation. Link the documents from
- Short answer: Yes. Both can distribute reference documents alongside a skill. A small