{"skill":{"slug":"codex-ppt","displayName":"Codex PPT","summary":"Generate image-based PowerPoint decks from articles, reports, papers, notes, or outlines. Use this skill when the user asks to create a visually unified PPT/...","tags":{"latest":"0.3.0"},"stats":{"comments":0,"downloads":215,"installsAllTime":0,"installsCurrent":0,"stars":1,"versions":4},"createdAt":1778337851802,"updatedAt":1778530306805},"latestVersion":{"version":"0.3.0","createdAt":1778528622169,"changelog":"### Documentation\n\n- Add an `Attention Is All You Need` example deck with preserved paper figures. (#18)\n- Require subagent-based per-slide generation whenever subagents are available. (#18)\n- Add mandatory phase gates and stricter image backend selection rules to the codex-ppt skill. (#18)\n- Add README examples for assigning original paper figures to specific slides, referencing issue #14. (#18)\n- Document Markdown image references inside `Required images`. (#18)\n- Clarify how to preserve user-supplied figures and result charts in generated decks, including subagent slide generation with approved sample-slide style references. (#18)\n- Document simplified prompt preparation with one self-contained JSON job per slide instead of a separate prompt file and job manifest. (#18)\n- Document relative input image path resolution in slide jobs against the `deck_spec.json` directory. (#18)\n- Reject duplicate slide numbers when preparing per-slide prompt jobs. (#18)\n- Clarify that CLI/API fallback must not text-generate slide jobs that require input images. (#18)","license":"MIT-0"},"metadata":{"os":null,"systems":null},"owner":{"handle":"ningzimu","userId":"s17e4dp91njzkhm04gy32mxzah83g9sk","displayName":"凝紫暮","image":"https://avatars.githubusercontent.com/u/33387665?v=4"},"moderation":null}