菜单
Build a library of metacognitive prompts targeting planning, monitoring, or evaluation for a specific task. Use when developing students' thinking-about-thinking during independent work.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/GarethManning/education-agent-skills --skill metacognitive-prompt-library命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
基于 SOC 职业分类
| name | metacognitive-prompt-library |
| description | Build a library of metacognitive prompts targeting planning, monitoring, or evaluation for a specific task. Use when developing students' thinking-about-thinking during independent work. |
| disable-model-invocation | false |
| user-invocable | true |
| effort | medium |
| skill_id | self-regulated-learning/metacognitive-prompt-library |
| skill_name | Metacognitive Prompt Library |
| domain | self-regulated-learning |
| version | 1.0 |
| evidence_strength | strong |
| evidence_sources | ["Flavell (1979) — Metacognition and cognitive monitoring: a new area of cognitive-developmental inquiry","Veenman et al. (2006) — Metacognition and learning: conceptual and methodological considerations","Hattie (2009) — Visible Learning: metacognitive strategies effect size ~0.69","Schraw (1998) — Promoting general metacognitive awareness","Tanner (2012) — Promoting student metacognition in the sciences"] |
| input_schema | {"required":[{"field":"task_description","type":"string","description":"The specific learning task students are completing"},{"field":"student_level","type":"string","description":"Age/year group and metacognitive development level"},{"field":"metacognitive_focus","type":"string","description":"Which metacognitive skill to target: monitoring, evaluating, planning, or debugging"}],"optional":[{"field":"subject_area","type":"string","description":"Subject context for domain-specific metacognitive prompts"},{"field":"student_profiles","type":"array","description":"From context engine: individual metacognitive profiles, common stuck points"},{"field":"task_phase","type":"string","description":"When prompts will be used: before, during, or after the task"}]} |
| output_schema | {"type":"object","fields":[{"field":"prompts","type":"array","description":"Ordered set of metacognitive prompts with type labels and usage timing"},{"field":"teacher_script","type":"string","description":"How to introduce and model metacognitive prompting"},{"field":"student_card","type":"string","description":"Copy-pasteable prompt card for student use"}]} |
| chains_well_with | ["self-regulation-scaffold-generator","study-strategy-selector","error-analysis-protocol","elaborative-interrogation-generator"] |
| teacher_time | 3 minutes |
| tags | ["metacognition","monitoring","thinking-about-thinking","self-regulation","comprehension"] |
Generates a calibrated set of metacognitive prompts for a specific learning task — prompts that help students monitor their own comprehension, evaluate their strategy effectiveness, detect errors, and adjust their approach. Unlike generic "think about your thinking" instructions, these prompts are task-specific and targeted at a named metacognitive skill (monitoring, evaluating, planning, or debugging). AI is specifically valuable here because effective metacognitive prompts must sit in a precise zone: concrete enough that students know what to do, but open enough that they genuinely reflect rather than just comply. Most teacher-generated reflection prompts are either too vague ("How did it go?") or too directive ("Did you use a topic sentence?"), missing the metacognitive level entirely.
Flavell (1979) defined metacognition as "knowledge and cognition about cognitive phenomena" — thinking about thinking. He distinguished metacognitive knowledge (what you know about your own cognition) from metacognitive regulation (how you control it). Veenman et al. (2006) demonstrated that metacognitive skilfulness is a stronger predictor of learning outcomes than intelligence, accounting for approximately 17% of variance in academic performance even after controlling for IQ. Hattie's (2009) synthesis found metacognitive strategies produce an effect size of approximately 0.69. Schraw (1998) identified that metacognitive awareness can be explicitly taught through structured prompting — students who regularly use metacognitive prompts develop stronger monitoring skills over time. Tanner (2012) applied metacognitive research specifically to science education, showing that discipline-specific metacognitive prompts (e.g., "What evidence would change my conclusion?") are more effective than generic ones. The evidence is clear: metacognition is teachable, transferable, and among the highest-leverage interventions available.
The teacher must provide:
Optional (injected by context engine if available):
You are an expert in metacognition research, specialising in Flavell's (1979) framework, Veenman et al.'s (2006) work on metacognitive skilfulness, and the practical application of metacognitive prompting in classroom settings (Tanner, 2012; Schraw, 1998).
Your task is to generate a set of metacognitive prompts for the following:
**Task:** {{task_description}}
**Student level:** {{student_level}}
**Metacognitive focus:** {{metacognitive_focus}}
The following optional context may or may not be provided. Use whatever is available; ignore any fields marked "not provided."
**Subject area:** {{subject_area}} — if not provided, infer from the task description and generate domain-appropriate prompts.
**Student profiles:** {{student_profiles}} — if not provided, design for a typical class at the stated metacognitive development level.
**Task phase:** {{task_phase}} — if not provided, generate prompts for all three phases (before, during, after), weighted toward the phase most relevant to the metacognitive focus.
Apply these evidence-based principles:
1. **Distinguish metacognitive types (Flavell, 1979; Schraw, 1998):**
- **Monitoring prompts:** Help students assess their current understanding in real time. "Do I actually understand this, or does it just feel familiar?" These target the illusion of knowing — the gap between feeling you understand and being able to explain or apply.
- **Evaluating prompts:** Help students judge the quality of their work or strategy against criteria. "Is my answer reasonable? How would I check?"
- **Planning prompts:** Help students select and sequence strategies before starting. "What approach will work here? What's my first step?"
- **Debugging prompts:** Help students identify and fix breakdowns in understanding. "Where exactly did I get lost? What was the last thing I understood?"
2. **Make prompts task-specific (Tanner, 2012):** Every prompt must reference the actual task. "Am I understanding this?" is not metacognitive — it's too vague to act on. "Can I explain in my own words why the character made that decision, without looking at the text?" is metacognitive because it provides a concrete test of comprehension.
3. **Calibrate to metacognitive maturity:**
- **Low metacognitive awareness:** Use concrete, closed prompts with observable actions. "Cover the text. Can you write down the three main causes from memory? If you can't, you don't know it yet — go back and re-read."
- **Moderate metacognitive awareness:** Use open prompts that require judgment. "How confident are you that your method is correct? What would you do to check?"
- **High metacognitive awareness:** Use reflective prompts that target strategic thinking. "Is this the most efficient strategy, or am I using it because it's familiar?"
4. **Target the illusion of knowing (Veenman et al., 2006):** The most common metacognitive failure is believing you understand something when you can't actually apply it. The most valuable prompts expose this gap — they force students to demonstrate understanding (explain, apply, predict) rather than simply rate it ("I get it").
5. **Prompt, don't instruct:** Metacognitive prompts ask questions that trigger reflection. They do not tell students what to think. "Check your answer" is an instruction. "What would you expect the answer to be roughly, before you calculate? Does your actual answer match that expectation?" is a metacognitive prompt.
Return your output in this exact format:
## Metacognitive Prompts: [Task Name]
**For:** [Student level]
**Focus:** [Metacognitive skill being targeted]
### Before the Task (Planning)
[3–4 prompts that activate metacognitive planning]
### During the Task (Monitoring)
[4–6 prompts calibrated to key moments in the task where metacognitive monitoring is most needed]
### After the Task (Evaluating)
[3–4 prompts for post-task metacognitive evaluation]
### Teacher Modelling Script
[A brief script showing how the teacher models metacognitive thinking aloud for one prompt — making the invisible visible]
### Student Prompt Card
[A clean, copy-pasteable card with 5–6 key prompts in student-friendly language]
**Self-check before returning output:** Verify that (a) every prompt is specific to this task, (b) prompts ask questions rather than give instructions, (c) at least two prompts specifically target the illusion of knowing, (d) prompts are calibrated to the stated metacognitive maturity level, and (e) the student card uses language appropriate for the age group.
Scenario: Task: "Solving multi-step word problems in Year 9 Mathematics (algebra and ratio)" / Student level: "Year 9, low metacognitive awareness — students often say 'I get it' then can't do the problems independently" / Metacognitive focus: "Monitoring — targeting the illusion of knowing"
For: Year 9 Mathematics, low metacognitive awareness Focus: Monitoring — specifically targeting the gap between "I get it" and being able to do it
P1. Read the word problem once without touching your pen. Now close the problem. Can you describe what the problem is asking in your own words? If you can't, read it again — but this time, underline the question it's actually asking you to answer.
P2. Before you calculate anything: is this problem about algebra, ratio, or both? How do you know? What clue in the wording told you?
P3. What is your first step going to be? Don't say "solve it" — what specifically will you write down first? If you're not sure, look at the problem and find one piece of information you can work with.
M1. You've just completed a step. Stop. Can you explain to an invisible partner why you did that step? If you can say "because..." and the reason makes sense, continue. If you can only say "because that's what you do," you may be following a procedure without understanding it — and that means you'll get stuck on the next unfamiliar problem.
M2. Before you move to the next step, estimate: roughly what size should the answer be? If you're calculating someone's age, it should be between 0 and 120. If you're calculating a ratio of lengths, it should be positive. Write your estimate in the margin. When you get your answer, check it against this estimate.
M3. You're stuck. That's normal. Before asking for help, answer these questions:
M4. You've got an answer. Now do this test: put your hand over your working. Can you explain to someone how you got from the problem to the answer, step by step, without looking? If not, your understanding is shallower than you think.
M5. Look at your answer. Does it make sense in the real-world context of the problem? If the problem asks how many buses are needed for 200 students and you got 6.67 buses, something needs adjusting. If a student's test score is negative, something went wrong. Context is your error-detection system.
E1. Rate your confidence in your answer: 1 (guessing), 2 (think so but not sure), 3 (confident), 4 (certain and could explain to someone else). Now — if you rated yourself a 3 or 4, prove it: explain your method in two sentences without looking at your working.
E2. Think about where you got stuck (if you did). Was it because: (a) I didn't understand what the problem was asking → reading/comprehension issue (b) I didn't know which method to use → strategy selection issue (c) I knew the method but made a calculation error → execution issue (d) I didn't get stuck — it went smoothly → great, but check M4 to make sure
This matters because each type of stuck needs a different fix.
E3. Compare your approach to a partner's. Did you use the same method? If different, which was more efficient? Could you have done it their way? What does this tell you about how many methods you have available for this type of problem?
"I'm going to show you what goes on inside my head when I solve a word problem. Watch what I do — I'm going to think out loud.
[Reads problem aloud] 'A school trip costs £12 per student plus a £45 bus hire. 28 students go on the trip. The school has a budget of £400. Is the budget enough?'
OK, first thing — I'm not going to start calculating yet. Let me re-read and ask myself: what is this problem actually asking? It's asking whether £400 is enough. So my answer isn't a calculation — it's a yes or no, backed up by numbers.
Now, what's my first step? I need to work out the total cost. That's 28 times £12, plus £45. Let me estimate first — 28 times 12 is roughly 30 times 12, which is £360, plus £45 is about £405. So it's going to be close. That estimate will help me check my exact answer.
[Calculates] 28 × 12 = 336. Plus 45 = £381. Is £381 less than £400? Yes. So the budget is enough, with £19 to spare.
Now — here's the metacognitive bit — let me check. Does my answer make sense? £381 for 28 students on a trip — that's about £13.60 per student. Does that feel reasonable? Yes. And my exact answer (£381) is close to my estimate (£405), so I haven't made a major calculation error.
That process — estimating first, checking against context, asking whether my answer makes sense — that's what I want you to do every time."
WORD PROBLEM THINKING CHECKLIST
Before I start:
While I'm working:
When I finish:
Metacognitive prompts require teacher modelling to be effective. Distributing a prompt card without explicitly modelling what metacognitive monitoring looks and sounds like will produce compliance (students ticking boxes) rather than genuine metacognition. Tanner (2012) is clear: the teacher must think aloud first, making invisible cognitive processes visible, before students can internalise the prompts.
Low metacognitive awareness students may find open-ended monitoring prompts frustrating or confusing. For students who have never been asked "Can you explain why you did that step?", the initial experience is uncomfortable. Start with the most concrete prompts (M2: estimate before calculating) and gradually introduce more reflective ones. Do not launch all prompts simultaneously.
Metacognitive monitoring takes time and initially slows task completion. Teachers under time pressure may be tempted to skip the monitoring checkpoints. This is counterproductive — the monitoring IS the learning intervention. If time is short, use fewer problems with monitoring rather than more problems without it. Quality of processing beats quantity of practice.