Practical AI Skills

中文说明见 README.zh-CN.md.

Two reusable Codex skills built from everyday study and tool-selection pain points:

• evaluate-ai-tools: compare AI tools with real tasks, scorecards, and evidence.
• compress-study-materials: turn long notes, PDFs, articles, lectures, and reports into compact study outputs.

Why This Exists

I kept running into two repeated problems while learning and trying AI products:

1. There are too many AI tools, but casual impressions are noisy. I needed a fair way to compare tools against my real workflow.
2. Learning materials pile up quickly. Summaries are easy to generate, but it is harder to turn material into notes, self-tests, Anki cards, and review plans.

This repository turns those repeated workflows into reusable skills that other students, job seekers, researchers, and AI tool users can adapt.

What's Inside

Skill	Use it when	Main outputs
evaluate-ai-tools	You need to compare AI tools, models, copilots, writing assistants, search tools, or productivity apps	Task suite, scoring rubric, evidence log, weighted scorecard, recommendation
compress-study-materials	You need to compress long learning materials into something reviewable	Study pack, concept map, self-test, Anki table, reading log, review plan

Quick Install

Copy the skill folders into your Codex skills directory:

mkdir -p ~/.codex/skills
cp -R skills/evaluate-ai-tools ~/.codex/skills/
cp -R skills/compress-study-materials ~/.codex/skills/

Then invoke them in Codex:

Use $evaluate-ai-tools to compare these three AI writing tools for my weekly research workflow.

Use $compress-study-materials to turn this lecture note into a study pack with Anki cards.

Example Outputs

• AI tool evaluation report
• Study compression output
• Sample scoring CSV

Utility Scripts

Aggregate AI tool scores:

python3 skills/evaluate-ai-tools/scripts/score_tools.py examples/sample-ai-tool-scores.csv --out /tmp/ai-tool-scorecard.md

Export Anki cards from a Markdown study pack:

python3 skills/compress-study-materials/scripts/anki_from_markdown.py examples/study-compression-output.md --out /tmp/anki-cards.csv

Add --tagged-only to export only rows that include a Tags value.

Run repository checks:

python3 scripts/validate_repo.py

Design Principles

• Start from real user tasks, not generic prompts.
• Separate observed evidence from marketing claims.
• Make outputs reusable by someone who did not run the workflow.
• Prefer active recall and decision-ready summaries over long passive notes.
• Keep each skill small enough to load quickly, with deeper details in references/.

Roadmap

• Add more example evaluations for study, writing, research, and job-search workflows.
• Add bilingual templates for Chinese/English study outputs.
• Add optional scoring presets for students, content creators, and business teams.
• Add more deterministic helpers for cleaning card tables and comparing scorecards over time.

If this saves you time, a star helps more people find it.