Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create companion Codex workflows from book chapters, reader exercises, prompts, field notes, and teaching material. Use when Codex needs to turn a chapter excerpt, 書籍練習, AI 提示詞, worksheet, or author note into a reusable Skill-style workflow, GitHub-shareable companion artifact, or book link resource for readers.
.claude/skills/twhsi-book-companion-skill/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 136% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 65% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 6% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -6% | 0% |
Convert book material into a reader-friendly, executable Codex workflow. Preserve the author's voice, make the task usable by someone who only has the book in front of them, and produce artifacts that can be shared through GitHub or linked from a printed/e-book chapter.
SKILL.md workflow for repeatable use.Use this shape when the reader should copy one prompt from GitHub or a QR page:
markdown## Chapter X: <exercise name> Paste this into Codex: > Use this chapter exercise to help me <reader goal>. My raw material is below. Ask at most two clarifying questions if needed, then produce <expected output>.
Use this shape when the book material should become a reusable Codex Skill:
markdown--- name: <short-hyphen-name> description: <what the skill does and exact trigger scenarios> --- # <Human Title> ## Overview <one or two sentences> ## Workflow <steps another Codex instance can follow>
Use this shape for GitHub pages, README sections, and book links:
markdown## <Chapter / Tool / Exercise Name> **Use when:** <reader situation> **Input:** <what the reader brings> **Codex prompt:** `<single concise prompt>` **Output:** <what Codex returns>
description frontmatter, not only in the body.SKILL.md lean; move long examples or chapter-specific details to references/.Read references/chapter-workflow-patterns.md when converting multiple chapter exercises, building a companion README, or standardizing many book links.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 7,776 | 8,362 | +8% | 1 | 1 | 0% | 1,189 | 1,682 | +41% | 0 | 0 | — |
case-02 | fail→pass | 35,788 | 11,174 | -69% | 1 | 1 | 0% | 1,084 | 2,559 | +136% | 0 | 0 | — |
case-03 | fail→pass | 6,852 | 6,247 | -9% | 1 | 1 | 0% | 1,078 | 1,776 | +65% | 0 | 0 | — |
case-04 | fail→pass | 10,841 | 6,279 | -42% | 1 | 1 | 0% | 1,562 | 1,649 | +6% | 0 | 0 | — |
case-05 | pass→pass | 17,731 | 8,098 | -54% | 1 | 1 | 0% | 2,980 | 2,012 | -32% | 0 | 0 | — |
case-06 | fail→pass | 16,954 | 8,793 | -48% | 1 | 1 | 0% | 2,636 | 2,059 | -22% | 0 | 0 | — |
case-07 | pass→pass | 12,918 | 7,638 | -41% | 1 | 1 | 0% | 1,830 | 1,857 | +1% | 0 | 0 | — |
case-08 | pass→pass | 11,506 | 7,331 | -36% | 1 | 1 | 0% | 1,620 | 1,753 | +8% | 0 | 0 | — |
case-09 | fail→pass | 15,153 | 9,931 | -34% | 1 | 1 | 0% | 2,191 | 2,070 | -6% | 0 | 0 | — |
case-10 | pass→pass | 7,598 | 4,903 | -35% | 1 | 1 | 0% | 1,164 | 1,515 | +30% | 0 | 0 | — |
case-11 | fail→pass | 7,285 | 8,481 | +16% | 1 | 1 | 0% | 963 | 2,026 | +110% | 0 | 0 | — |
case-12 | fail→pass | 9,394 | 4,276 | -54% | 1 | 1 | 0% | 1,491 | 1,378 | -8% | 0 | 0 | — |
case-13 | pass→pass | 12,208 | 7,970 | -35% | 1 | 1 | 0% | 1,715 | 1,878 | +10% | 0 | 0 | — |
case-14 | pass→pass | 10,065 | 5,046 | -50% | 1 | 1 | 0% | 1,476 | 1,534 | +4% | 0 | 0 | — |
case-15 | pass→pass | 7,040 | 3,923 | -44% | 1 | 1 | 0% | 1,000 | 1,290 | +29% | 0 | 0 | — |
case-16 | pass→pass | 10,413 | 6,403 | -39% | 1 | 1 | 0% | 1,599 | 1,681 | +5% | 0 | 0 | — |
case-17 | pass→pass | 15,784 | 11,841 | -25% | 1 | 1 | 0% | 2,426 | 2,530 | +4% | 0 | 0 | — |
case-18 | pass→pass | 10,716 | 5,411 | -50% | 1 | 1 | 0% | 1,498 | 1,552 | +4% | 0 | 0 | — |
case-19 | pass→pass | 7,691 | 3,963 | -48% | 1 | 1 | 0% | 1,137 | 1,302 | +15% | 0 | 0 | — |
case-20 | fail→fail | 2,573 | 5,811 | +126% | 1 | 1 | 0% | 351 | 1,538 | +338% | 0 | 0 | — |
case-21 | pass→pass | 7,302 | 5,323 | -27% | 1 | 1 | 0% | 1,306 | 1,669 | +28% | 0 | 0 | — |
case-22 | pass→pass | 16,261 | 13,195 | -19% | 1 | 1 | 0% | 2,630 | 2,767 | +5% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +32 percentage points is the difference between those two pass rates over the 21 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.