Interactive Study Cards

简体中文

Interactive Study Cards is a Codex skill for turning user-provided notes, concepts, code snippets, formulas, or learning material into a single self-contained HTML study page.

The generated pages are designed for active recall, not passive reading. Cards are source-grounded, atomic, mobile-friendly, and can include MathML formulas, stable visual explanations, weak-card review, keyboard shortcuts, and filtering depending on the selected output level. The visual system is cross-disciplinary: it chooses templates by knowledge structure, not by subject.

Features

• Creates one complete HTML file with inline CSS and JavaScript.
• Uses source-locked generation by default to avoid unsupported facts.
• Tracks source support levels and source-audit evidence so scanned or range-only materials do not get over-precise page claims.
• Builds active recall cards from notes, formulas, code, comparisons, and mechanisms.
• Supports Mini, Pro, and Max output levels.
• Uses MathML for formulas when possible.
• Adds formula and method caveats for math/science topics, including domains, hypotheses, nonzero divisors, and lost-solution risks.
• Includes a cross-disciplinary visual template library for definitions, comparisons, classifications, mechanisms, cause-effect chains, variable relationships, systems, arguments, language patterns, and state traces.
• Uses stable HTML/CSS learning visuals first, such as trace tables, state tables, comparison grids, code/output panels, and before/after panels.
• Uses inline SVG only for template-bound diagrams that need vector geometry.
• Avoids external libraries, CDNs, MathJax, KaTeX, and external assets unless explicitly allowed.
• Supports weak-card marking and local review memory in Pro and Max modes.
• Supports search, tag filters, shuffle, stats, and weak-card JSON export in Max mode.

Install

Quick install or update:

curl -fsSL https://raw.githubusercontent.com/Lorlds/interactive-study-cards/main/install.sh | bash

Install a specific release:

curl -fsSL https://raw.githubusercontent.com/Lorlds/interactive-study-cards/main/install.sh | bash -s -- v1.0.4.040726

Manual install into Codex's skill discovery directory:

git clone https://github.com/Lorlds/interactive-study-cards.git ~/.codex/skills/interactive-study-cards

If you already cloned it elsewhere, link it instead:

mkdir -p ~/.codex/skills
ln -s /path/to/interactive-study-cards ~/.codex/skills/interactive-study-cards

Restart Codex after installing so the new skill is loaded.

Releases

Versions start at v1.0.0.040726 and use the format vMAJOR.MINOR.PATCH.DDMMYY. If multiple releases happen on the same date, increment PATCH.

Every GitHub Release includes a short, explicit list of what changed.

Usage

Ask Codex for interactive study cards and provide the source material:

Use $interactive-study-cards to make Pro HTML study cards from these microeconomics notes:
...

You can also specify the output level:

Create Max interactive study cards from this calculus review sheet. Focus on graph interpretation and common mistakes.

If the request is missing important details, the skill asks for subject area, source material, target depth, output level, and any special focus.

Output Levels

Mini is the fastest simple review page. It includes one-card-at-a-time review, hidden answers, previous and next controls, progress, tags, mobile layout, formulas, and simple template visuals when useful.

Pro is the default. It adds random review, keyboard shortcuts, weak-card marking, localStorage weak-card memory, weak-only mode, and frequent visual explanations for visualizable concepts. It prefers stable HTML/CSS templates before SVG.

Max is the complete review tool. It adds tag filters, search, shuffle mode, reset progress, weak-card JSON export, session stats, and richer template-bound visual explanations such as trace tables, comparison grids, graphs, number lines, flowcharts, force diagrams, and process diagrams when supported by the source.

Repository Layout

.
├── LICENSE
├── SKILL.md
├── agents/
│   └── openai.yaml
├── install.sh
├── references/
│   ├── math-guardrails.md
│   └── visual-templates.md
├── README.zh-CN.md
└── README.md

Grounding

This skill treats the user's material as the primary source. It should not silently add unsupported claims, formulas, examples, mechanisms, visuals, or diagrams. If a claim is uncertain and verification is unavailable, the skill should omit it or ask for clarification.

For scanned or image-only files, exact page labels require OCR or page-image inspection. Range labels require explicit source-audit evidence such as a table of contents entry, bookmark, heading, page image, OCR heading, or user-provided page range. Otherwise, the skill should mark standard-topic reminders as prerequisite and avoid source-specific wording.

License

MIT. See LICENSE.