讓 AI 成為
你的教學拍檔。

---
name: lesson-plan-builder
description: Builds a complete single-lesson plan - measurable objective, timed four-part arc (hook, direct instruction, guided practice, closure), a formative check with a proceed/re-teach rule, and differentiation notes - clear enough that a substitute could run it. Use when someone asks "write a lesson plan for...", "plan a 50-minute class on photosynthesis", "how should I structure tomorrow's lesson", or "turn this standard into a lesson". Do NOT use for sequencing a multi-week unit or full course - use curriculum-mapper instead; for deep per-learner adaptation of an existing lesson, use differentiated-instruction; for writing the quiz itself, use quiz-generator.
metadata:
  title: "Lesson Plan Builder"
---

# Lesson Plan Builder

A strong lesson plan is a contract between the teacher and the learners: here is where we start, here is what we will do, here is how we will know we arrived. The costly failure this skill prevents is the lecture-heavy plan - the teacher talks for 30 minutes, practice gets squeezed into the last 5, and the first evidence of confusion arrives on the graded test, when it is too late to act.

## Operating procedure

Work the steps in order: the objective must exist before the exit task, and the exit task before the activities, because activities chosen first will smuggle in their own objectives.

### Step 1: Gather inputs

Collect before drafting. Where the user cannot answer, use the default and label the assumption a guess.

1. Subject and topic (required).
2. Grade or level (required - it sets vocabulary and pacing).
3. Class length in minutes (default: 50).
4. What students already know - the prior lesson or prerequisite skill (default: ask; if unavailable, state the assumed prerequisite as a guess).
5. Standards to align, if any (default: none, omit the alignment row).
6. Constraints: available materials, tech, class size, known student needs.

### Step 2: Write the objective first

Use the form: Students will be able to [observable verb] [specific content] [condition or standard]. Bloom's verbs sharpen the cognitive level: identify, compare, construct, evaluate. Reject vague verbs - "understand" and "appreciate" cannot be observed. One primary objective per lesson; secondary objectives are optional and labeled as secondary.

### Step 3: Design the exit task before any activity

Backward design: write the summative check or exit task now, mirroring the objective wording. If the objective says "compare two characters using textual evidence," the exit task asks students to compare two characters using textual evidence - not to list character traits.

### Step 4: Build the four-part arc with a time budget

For a 50-minute class (scale proportionally for other lengths):

- Hook: 3-7 minutes (5 is typical). Activate prior knowledge or pose a productive puzzle. If the hook runs past 7 minutes it is eating practice time.
- Direct instruction / modeling: under 15 minutes, hard ceiling. If the content cannot be modeled in 15 minutes, split it across two lessons rather than lecturing longer.
- Guided then independent practice: the majority of the lesson - at least half the total minutes (25+ of 50). Students doing, not watching.
- Closure: 3-5 minutes. Students synthesize ("what changed in your thinking?"), not just summarize.

### Step 5: Filter activities against the objective

For each candidate activity ask: if a student does this activity well, are they closer to meeting the objective? If no, cut it - engagement alone does not qualify. Prefer activities that generate visible evidence of learning: think-pair-share, exit tickets, mini-whiteboards.

### Step 6: Place the formative check with a decision rule

At least one mid-lesson check, positioned after guided practice begins and before independent practice. Name both signals numerically: proceed if roughly 80% or more of responses are correct; re-teach if below - and name the re-teach move (a second worked example, a peer explanation round), not just "review it."

### Step 7: Add materials and differentiation notes

List only materials actually required; flag anything needing advance preparation. One modification for students needing support, one extension for early finishers. Brief notes, not separate plans.

### Step 8: Run the substitute-teacher test

Read the plan as someone who has never met this class. Every segment must say what the teacher does and what students do. Fix anything that requires insider knowledge.

## Template

Copy and replace the [FILL] fields.

```
LESSON PLAN - [FILL: topic] - [FILL: grade] - [FILL: duration] min
Objective: Students will be able to [FILL: observable verb] [FILL: content] [FILL: condition/standard].
Standard: [FILL or omit]   Prerequisites assumed: [FILL]

FLOW
| Segment              | Time            | What happens                              | Materials |
| Hook                 | [FILL: 3-7] min | [FILL: puzzle or prior-knowledge prompt]  | [FILL]    |
| Direct instruction   | [FILL: <=15] min| [FILL: model + worked example]            | [FILL]    |
| Guided practice      | [FILL] min      | [FILL: students doing, teacher circulating]| [FILL]   |
| Independent practice | [FILL] min      | [FILL]                                    | [FILL]    |
| Closure              | [FILL: 3-5] min | [FILL: synthesis prompt]                  | [FILL]    |

FORMATIVE CHECK (during guided practice)
  Check:      [FILL: e.g. one mini-whiteboard problem]
  Proceed if: [FILL: e.g. >=80% correct]
  Re-teach if below - move: [FILL: e.g. second worked example, then re-check]

EXIT TASK (mirrors objective wording): [FILL]
DIFFERENTIATION - support: [FILL] | extension: [FILL]
ADVANCE PREP: [FILL]
```

## Deliverable

Produce a single scannable document containing: the objective, grade and duration, standards alignment (if provided), the flow table with per-segment times, the formative check with its proceed/re-teach rule, the exit task, and differentiation notes - using the template above.

## Do NOT

- Do not start from a favorite activity and reverse-engineer an objective; the objective comes first or the lesson drifts.
- Do not let direct instruction exceed 15 minutes - past that point retention drops and practice time disappears.
- Do not write a formative check without a numeric decision rule; "check for understanding" without a threshold produces no decision.
- Do not write an exit task that tests something other than the objective's verb (an objective about comparing, checked by a recall quiz, measures nothing).
- Do not plan a unit here - one lesson per plan; route multi-week sequencing to curriculum-mapper.

## Quality bar

- The objective uses an observable verb and one primary target.
- Practice time is at least half the lesson; instruction is under 15 minutes.
- The formative check names both a proceed threshold and a re-teach move.
- The exit task mirrors the objective wording.
- A substitute teacher could run the lesson without asking a question.

---

MIT License

Copyright (c) 2026 Alexander Ouellet

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

# 來源與授權

名稱：lesson-plan-builder
來源：https://github.com/SkillMedev/skills/blob/97a13607d669157a4a16d5ba79b77f208319f7b9/skills/lesson-plan-builder/SKILL.md
儲存庫：https://github.com/SkillMedev/skills
固定版本：97a13607d669157a4a16d5ba79b77f208319f7b9
授權：MIT（見同資料夾 LICENSE）
Copyright (c) 2026 Alexander Ouellet

SKILL.md 保留原作者英文內容，沒有翻譯或改寫。本站只補充這份來源記錄。
原資料夾只有 SKILL.md，沒有必需的腳本、外部服務或其他相依檔案。指示中提及其他 Skills 是不同任務的建議，不代表本下載包含那些 Skills。

本網站已核對檔案結構與原文一致性。可在網站分別複製名稱、說明與指示，在 Create manually 手動建立 Skill；也可選擇 Upload 匯入獨立 Skill ZIP，核對後按 Create。建立後須選用該 Skill，加入測試素材再驗證。

原文的教學比例與規則是作者的設計選擇，教師應按自己的學生、課程及實際測試調整。

---
name: quiz-generator
description: Writes quizzes and assessments with a stated Bloom's-level distribution, misconception-based distractors, and an answer key with rationales - plus a summary table of level, type, and points per question. Use when someone asks "write a quiz on...", "make a 10-question test for chapter 5", "generate review questions", or "check these multiple-choice questions for quality". Do NOT use for building the grading rubric for an essay or project - use rubric-builder instead; for designing the lesson the quiz sits inside, use lesson-plan-builder; for opinion or research surveys rather than knowledge assessment, use survey-designer.
metadata:
  title: "Quiz Generator"
---

# Quiz Generator

A well-constructed quiz is a learning event, not just a measuring event. The costly failure this skill prevents is the all-recall quiz with throwaway distractors: it reports high scores while misconceptions calcify undetected, and the teacher learns nothing actionable from the results.

## Operating procedure

### Step 1: Gather inputs

Collect before writing any question. Where the user cannot answer, use the default and label assumptions as guesses.

1. Topic and the specific objectives or content covered (required).
2. Grade or level (required).
3. Purpose: formative check, unit exam, or review/performance task (default: formative).
4. Question count and time budget (default: 10 questions; budget roughly 1 minute per multiple-choice item, 3-5 per short answer, 10+ per constructed response).
5. Known misconceptions in this topic (ask - these become distractors; if none supplied, derive them from typical partial understandings and label them as guesses).
6. Format constraints: paper vs digital, calculator allowed, open notes.

### Step 2: State the Bloom's distribution before writing questions

Bloom's six levels, lower to higher order: Remember, Understand, Apply, Analyze, Evaluate, Create. Declare the target mix first; do not drift toward all-recall by default.

- Formative quiz: covers Remember through Apply - roughly 40% Remember/Understand, 60% Apply.
- Unit exam: must reach Analyze and Evaluate - no more than 50% of points on Remember/Understand.
- Performance/review task: Evaluate and Create dominate.

Red line: if every question is Remember-level, the quiz measures memorization only - rebalance before proceeding.

### Step 3: Match question type to level

- Multiple choice: efficient for Remember and Understand; use sparingly for Apply and above.
- Short answer: versatile across Apply, Analyze, Evaluate.
- Constructed response / scenario-based: best for Analyze and above.
- True/false: low signal - diagnostic pre-checks only, never graded summative work.

### Step 4: Write stems, then distractors from misconceptions

Lower-order stems are direct and simple: define, identify, list. Higher-order stems use scenarios, data sets, or documents as stimulus - kept concise and free of culturally specific references that create construct-irrelevant difficulty.

Every multiple-choice distractor must be the answer a student with a specific partial understanding would choose. Rules: the correct answer must be unambiguously correct (if a second option survives a defensible interpretation, rewrite the stem, not the option); no "all of the above" / "none of the above" except in carefully controlled contexts; keep options parallel in length and grammar, since the longest option telegraphs the answer.

### Step 5: Build the answer key with rationales

For each question: the correct answer, a one-sentence rationale for why it is correct, and - for multiple choice - the misconception each distractor represents. This is part of the deliverable, not an extra.

### Step 6: Assemble and check the summary table

Group questions by cognitive level, labeled. Produce the table (question number, level, type, points) and verify the actual point distribution matches the Step 2 declaration. If it drifted, fix questions, not the declaration.

## Worked contrast: distractor quality

Topic: photosynthesis, Understand level.

**Bad:**
"What do plants need for photosynthesis? A) sunlight, water, CO2  B) rocks  C) music  D) darkness"
B, C, D are jokes no partially-informed student would pick; the item measures test-taking, not understanding.

**Good:**
"Where does most of the mass of a growing tree come from? A) carbon dioxide in the air (correct)  B) minerals absorbed from the soil  C) water pulled up by the roots  D) energy from sunlight converted to matter"
Distractor analysis: B is the classic "plants eat soil" misconception; C confuses the reactant present in largest volume with the mass source; D reflects the common energy-becomes-matter confusion. A student choosing B tells the teacher exactly what to re-teach.

## Deliverable

Produce: (1) the question set grouped and labeled by Bloom's level, with total points stated; (2) the summary table - question number, cognitive level, question type, point value; (3) the answer key with per-question rationale and per-distractor misconception notes. Offer a student-facing version (instructions, no key) and a teacher version as separate documents.

## Do NOT

- Do not default to all-recall; a quiz that never reaches Apply cannot distinguish understanding from memorization.
- Do not write filler distractors - every wrong option a student would never choose wastes an opportunity to detect a misconception.
- Do not use true/false on graded summative work; a coin flip scores 50%.
- Do not let one question test two things at once; a wrong answer becomes undiagnosable.
- Do not ship a key without rationales; "B" alone forces the teacher to re-derive every answer during grading.

## Quality bar

- The Bloom's distribution was declared before writing and the final table matches it.
- Every MC distractor maps to a named misconception; no joke options.
- Every correct answer survives the "could a well-prepared student defend a different option?" test.
- Stimulus material is concise and construct-relevant.
- Total points and per-question values are stated and sum correctly.

---

MIT License

Copyright (c) 2026 Alexander Ouellet

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

# 來源與授權

名稱：quiz-generator
來源：https://github.com/SkillMedev/skills/blob/97a13607d669157a4a16d5ba79b77f208319f7b9/skills/quiz-generator/SKILL.md
儲存庫：https://github.com/SkillMedev/skills
固定版本：97a13607d669157a4a16d5ba79b77f208319f7b9
授權：MIT（見同資料夾 LICENSE）
Copyright (c) 2026 Alexander Ouellet

SKILL.md 保留原作者英文內容，沒有翻譯或改寫。本站只補充這份來源記錄。
原資料夾只有 SKILL.md，沒有必需的腳本、外部服務或其他相依檔案。指示中提及其他 Skills 是不同任務的建議，不代表本下載包含那些 Skills。

本網站已核對檔案結構與原文一致性。可在網站分別複製名稱、說明與指示，在 Create manually 手動建立 Skill；也可選擇 Upload 匯入獨立 Skill ZIP，核對後按 Create。建立後須選用該 Skill，加入測試素材再驗證。

原文的教學比例與規則是作者的設計選擇，教師應按自己的學生、課程及實際測試調整。

---
name: rubric-builder
description: Builds grading rubrics - analytic, holistic, or single-point - with criteria traced to learning objectives, 3-5 performance levels, observable behavior in every cell, and defensible weighting. Use when someone asks "make a rubric for this essay", "how should I grade this project", "turn these objectives into scoring criteria", or "my rubric feels vague, fix it". Do NOT use for writing the quiz questions themselves - use quiz-generator instead; for rubrics that score job candidates in interviews, use screening-rubric-builder; for narrative comments on a specific student's work, use student-feedback-writer.
metadata:
  title: "Rubric Builder"
---

# Rubric Builder

A rubric does two jobs: it tells students exactly what mastery looks like before they begin, and it gives teachers a consistent, defensible standard when scoring. The costly failure this skill prevents is the adjective ladder - "excellent / good / fair / poor" - which students cannot act on before submitting and teachers cannot defend when a grade is challenged.

## Operating procedure

### Step 1: Gather inputs

1. The assignment or task being assessed (required - the actual prompt, not a summary).
2. The learning objectives it assesses (required; if the user has none written, draft them first and label them as guesses for confirmation).
3. Grade or level, and the stakes (minor credit vs major grade - this drives format).
4. Point total, if the gradebook requires one (default: 100).
5. Whether a student-facing version is needed (default: offer one).

### Step 2: Derive criteria from objectives

List the objectives, then derive one criterion per objective or per cluster of tightly related objectives. Trace test: if a criterion cannot be traced to a stated objective, cut it. Practical ceiling: 3-6 criteria; beyond 6, scoring time balloons and criteria start overlapping. Mechanics/conventions may earn one criterion at most, and only if the objectives include communication quality.

### Step 3: Choose the format

- Analytic (each criterion scored separately): default for any graded assignment worth more than minor credit; the only format that produces diagnostic feedback per criterion.
- Holistic (single overall score): quick checks and short, low-stakes tasks only.
- Single-point (describes proficient only, with narrative space either side): efficient for iterative draft work where the goal is revision, not a grade.

### Step 4: Set the performance levels

3-5 levels; four is the standard: Exceeds Expectations, Meets Expectations, Approaching Expectations, Does Not Yet Meet Expectations. Three levels work for simple tasks. Five is the ceiling - more levels create scoring drift between graders and are rarely worth the calibration effort. Label levels with descriptive language, not just numbers, so feedback is inherent in the score.

### Step 5: Write descriptors - proficient first, observable only

Write the Meets Expectations cell first for each criterion, then adjust up and down. Every cell must describe observable features of the work - what it contains, how it is structured, what is missing - never grader impressions. Banned words in cells: excellent, good, poor, adequate, some, few, needs improvement, appropriate. Cell test: could two graders who have never met apply this cell to the same paper and land on the same level? Adjacent levels must differ by a nameable feature, not an adverb ("cites two or more sources" vs "cites one source", not "cites well" vs "cites adequately").

### Step 6: Weight the criteria

Assign points or percentages reflecting each criterion's importance to the objectives - the central skill should carry the most weight (typically 30-40% of total), mechanics rarely more than 10-15%. Document the weighting rationale in one sentence so it can be shared with students and defended to stakeholders.

### Step 7: Calibrate before use

Score one strong, one middling, and one weak sample (real or imagined) against the rubric. Any cell that forced a judgment call gets rewritten before the rubric ships.

## Skeleton: analytic rubric

Copy and replace the [FILL] fields.

```
RUBRIC - [FILL: assignment] - [FILL: total points]
Assesses objectives: [FILL: list]
Weighting rationale: [FILL: one sentence]

| Criterion (weight)        | Exceeds (pts)          | Meets (pts)            | Approaching (pts)      | Not Yet (pts)          |
| [FILL: criterion 1] ([FILL]%) | [FILL: observable features beyond proficient] | [FILL: observable features of proficiency - write this cell FIRST] | [FILL: named gap vs Meets] | [FILL: what is absent] |
| [FILL: criterion 2] ([FILL]%) | [FILL]             | [FILL]                 | [FILL]                 | [FILL]                 |
| [FILL: criterion 3] ([FILL]%) | [FILL]             | [FILL]                 | [FILL]                 | [FILL]                 |
| TOTAL                     |                        |                        |                        | [FILL] pts             |
```

Example of a filled cell pair (essay, criterion "Use of evidence", Meets vs Approaching):
- Meets: "Every body paragraph includes at least one quoted or paraphrased source, cited, and followed by a sentence explaining how it supports the claim."
- Approaching: "Most body paragraphs include a source, but at least one quotation is dropped in without an explanation connecting it to the claim."

## Deliverable

Produce the rubric as a table - rows are criteria, columns are performance levels - with point values per cell, a total-points row, a header naming the assignment and the objectives assessed, and the one-sentence weighting rationale. Offer a student-facing plain-language version.

## Do NOT

- Do not write adjective ladders; "excellent vs good" ranks work without describing it, so students cannot self-assess and grades cannot be defended.
- Do not include a criterion untraceable to an objective; it grades something never taught.
- Do not exceed five levels; fine gradations feel rigorous but produce inter-grader drift.
- Do not let mechanics outweigh the target skill; a brilliant argument with comma errors should not score below a vacuous clean one.
- Do not use holistic format for major grades; a single number gives the student nothing to fix.

## Quality bar

- Every criterion traces to a stated objective; 3-6 criteria total.
- Every cell passes the two-grader test and names observable features.
- Adjacent levels differ by a nameable feature, not an adverb.
- Weights sum correctly and the heaviest weight sits on the central skill.
- The rubric was calibrated against at least three sample performances.

---

MIT License

Copyright (c) 2026 Alexander Ouellet

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

# 來源與授權

名稱：rubric-builder
來源：https://github.com/SkillMedev/skills/blob/97a13607d669157a4a16d5ba79b77f208319f7b9/skills/rubric-builder/SKILL.md
儲存庫：https://github.com/SkillMedev/skills
固定版本：97a13607d669157a4a16d5ba79b77f208319f7b9
授權：MIT（見同資料夾 LICENSE）
Copyright (c) 2026 Alexander Ouellet

SKILL.md 保留原作者英文內容，沒有翻譯或改寫。本站只補充這份來源記錄。
原資料夾只有 SKILL.md，沒有必需的腳本、外部服務或其他相依檔案。指示中提及其他 Skills 是不同任務的建議，不代表本下載包含那些 Skills。

本網站已核對檔案結構與原文一致性。可在網站分別複製名稱、說明與指示，在 Create manually 手動建立 Skill；也可選擇 Upload 匯入獨立 Skill ZIP，核對後按 Create。建立後須選用該 Skill，加入測試素材再驗證。

原文的教學比例與規則是作者的設計選擇，教師應按自己的學生、課程及實際測試調整。