Codex App Explain Skill

Portable layered teaching and re-explanation for Codex App.

This repository ships one installable Codex App skill for explanation work: concept teaching, derivation unpacking, slide or PDF passage explanation, and bounded homework-step explanation.

This public repository now also documents the stable mode split used for these tasks:

• short-concept mode for one concept or one local blocker
• prerequisite-chain mode for hidden dependency chains
• paper-sentence mode for explicit sentence-by-sentence reading
• full-layered mode for full rebuilds and derivations

In paper-sentence mode, quoted source text may receive minimal Unicode-first readability cleanup for OCR noise or poor ASCII-style symbol rendering. The cleanup is display-only: preserve the original wording and meaning, and do not turn the quote into paraphrase or explanation.

What Ships

• installable skill: codex-app-explain
• bundled references for teaching principles, source binding, and route adaptation
• a public mode reference for concept, chain, sentence-reading, and layered teaching routes
• root-level docs for installation, scope, and privacy boundary

Install / Use

• Codex App: install the skill from this repo path codex-app-explain
• GitHub install target:
  • repo: <owner>/codex-app-explain-skill
  • path: codex-app-explain
• Restart Codex App after installation so the new skill is discovered.

Trigger Examples

• Explain this definition step by step.
• Where does this derivation step come from?
• Teach me this equation from basic to rigorous.

Non-Trigger Examples

• Remember my formatting preference.
• Find this file for me.
• Write the full assignment for me.

Privacy Boundary

This public skill keeps the workflow generic and reusable.

• It excludes personal memory files, local profile data, and private companion workflows.
• The published source embeds no personal absolute paths.
• It keeps examples generic rather than course-specific or identity-specific.

Companion Boundary

This repository is the teaching-workflow half of the stack.

• It owns route selection, source binding, prerequisite handling, and stop conditions.
• It does not own persistent user memory or user-specific formatting.
• When a host environment provides a separate output-rules skill, that companion should own notation and layout while this skill continues to own the teaching route.
• Quoted-source readability cleanup in sentence-reading belongs here as part of the reading workflow; a separate rendering-rules companion may still publish the narrower symbol-cleanup subset for display consistency.

Repository Layout

• codex-app-explain/: installable Codex App skill
• codex-app-explain/references/: bundled public teaching references
• CHANGELOG.md: release history
• LICENSE: MIT

Chinese:

• README.zh-CN.md