Skip to content

GarethManning/education-agent-skills / skills/professional-learning/technological-pedagogical-content-knowledge-developer

technological-pedagogical-content-knowledge-developer

Verified

Develop TPACK for integrating a specific technology or AI tool into subject teaching with pedagogical alignment. Use when adopting new ed-tech, reviewing AI tools, or planning technology integration.

FileKindSize
SKILL.mdskill md26.1 KB
references/worked-examples.mdmarkdown18.4 KB

references/worked-examples.md

View on GitHub
## Example Output

**Scenario:** *Teaching context: "Year 9 History — causes of World War One, including source analysis and historical argumentation" / Technology in use: "Claude (AI assistant) — students will use Claude to help analyse primary sources and construct historical arguments about the Fischer thesis vs revisionist interpretation" / Learner stage: "14-15 years" / Teacher background: "Experienced history teacher with strong PCK for WWI topic, moderate technology confidence, first time using AI tools in the classroom" / Intended learning outcome: "Students develop historical thinking skills including source evaluation, evidence-based argumentation, and evaluation of competing historical interpretations"*

---

## TPACK Development Plan: Claude (AI Assistant) in Causes of WWI

**Teaching context:** Year 9 History — causes of WWI, source analysis, historical argumentation (Fischer vs revisionist)
**Technology:** Claude (AI assistant)
**Learner stage:** 14-15 years (Year 9)
**Teacher background:** Experienced history teacher, strong WWI PCK, moderate technology confidence, first time using AI in the classroom
**Intended learning outcome:** Historical thinking skills — source evaluation, evidence-based argumentation, evaluation of competing interpretations

### 1. TPACK Diagnosis

**Technology-Content Knowledge (TCK):**
The teacher likely does not know how Claude represents historical content. Key gaps:
- Does Claude reliably distinguish between Fischer's thesis and the revisionist position with historiographical accuracy? (It generally does for well-established historiographical debates, but may present synthesised summaries that smooth over the sharp disagreements between historians.)
- Can Claude accurately quote from primary sources? (No — large language models frequently fabricate or misattribute specific quotes and dates. The teacher must understand that Claude's paraphrases of primary sources are generally reliable but specific quotations must be verified.)
- Does Claude represent historical causation with appropriate complexity? (It tends toward balanced, diplomatic summaries rather than the sharp argumentative positions that characterise genuine historiographical debate. This may flatten the very debate the task is designed to explore.)

**Technology-Pedagogy Knowledge (TPK):**
The teacher likely has strong general pedagogical knowledge but does not know how AI tools interact with historical thinking pedagogy. Key gaps:
- The teacher may not recognise that using Claude to generate historical arguments removes the productive struggle with evidence that IS the core historical thinking skill. If students ask Claude "argue the Fischer thesis," they receive a fluent argument without doing the historical work of selecting evidence, weighing its significance, and constructing the argument themselves.
- The teacher may not know how to design an AI-integrated task that preserves the thinking demand — specifically, the difference between students using AI to do the thinking and students using AI to scaffold their own thinking.

**Full TPACK:**
The teacher has not yet used AI in the classroom and therefore lacks the experiential TPACK that comes from observing how students actually use the tool. They will not yet be able to make the real-time judgment: "this student is using Claude to think more deeply about the evidence" versus "this student is using Claude to avoid thinking about the evidence." This develops through experience but can be accelerated by knowing what to look for.

**TPACK strengths:**
- Strong PCK for WWI content — the teacher knows the historiography, the primary sources, and the student misconceptions. This is a significant advantage: they can evaluate whether Claude's historical content is accurate, which a teacher without this PCK could not.
- Strong pedagogical knowledge for historical thinking — the teacher knows how to design tasks that develop source evaluation and argumentation. The challenge is adapting these skills to an AI-mediated environment, not building them from scratch.

### 2. Technology-Content Analysis

**What Claude makes visible:**
- Claude can generate well-structured summaries of historiographical positions (Fischer, revisionist, post-revisionist) quickly, allowing students to access the debate faster than through reading alone. This is useful for providing context before students engage with primary sources.
- Claude can model the structure of a historical argument (claim → evidence → reasoning → qualification), giving students an example to compare their own work against.
- Claude can generate follow-up questions that push students to consider evidence they may have overlooked: "Have you considered the role of Russia's mobilisation in the Fischer debate?" This scaffolds the thinking without doing it.

**What Claude obscures or distorts:**
- Claude tends to present balanced, diplomatic syntheses of debates rather than the sharp, committed positions that characterise genuine historiography. Fischer's argument was not balanced — it was a provocative, evidence-based provocation that divided the historical profession. If Claude smooths this into "some historians blame Germany more, while others emphasise shared responsibility," it has distorted the intellectual force of the debate.
- Claude cannot evaluate the provenance of a primary source the way a historian does. It can describe what a source says but struggles to make the inferential leap: "The Kaiser's telegram to the Tsar may be sincere diplomacy or calculated performance — we cannot determine intent from the text alone." This is precisely the kind of historical thinking the task is designed to develop.
- Claude's "knowledge" of primary sources is derived from its training data, not from reading the original documents. It may present paraphrased versions of sources with subtle inaccuracies that a teacher without strong content knowledge would not catch.

**Reliability assessment:**
- **Reliable:** General summaries of the Fischer debate, the alliance system, the July Crisis sequence. These are well-established historical content that Claude handles accurately.
- **Approximately reliable:** Summaries of specific historians' positions. Claude captures the broad argument but may not represent the nuance or the specific evidence each historian relies on. Verify against the teacher's own knowledge.
- **Unreliable:** Specific primary source quotations, specific dates in the July Crisis (Claude may give slightly wrong dates), and claims about what specific documents "prove." The teacher must verify any specific factual claim Claude makes about primary sources before allowing it to be used as content with students.

### 3. Pedagogy-Technology Alignment

| Pedagogical move | Technology effect | Explanation |
|---|---|---|
| Source evaluation (provenance, reliability, utility) | **Undermines** | If students ask Claude to evaluate a source, they receive a competent evaluation without developing the skill of evaluation themselves. The source evaluation IS the thinking — outsourcing it to AI defeats the learning purpose. |
| Constructing evidence-based arguments | **Undermines** | If students use Claude to draft their Fischer-vs-revisionist argument, they produce a fluent product without doing the historical thinking the product is supposed to evidence. The argument construction IS the learning, not the argument itself. |
| Scaffolding argument structure | **Supports** | Claude can provide the structure of a historical argument (claim-evidence-reasoning-qualification) as a framework. Students can then fill the framework with their own content. This scaffolds without replacing the thinking. |
| Generating follow-up questions and counter-arguments | **Supports** | After a student has drafted their own argument, Claude can generate challenges: "A revisionist historian would respond to your claim by pointing to..." This pushes the student to engage with counter-evidence they may not have considered. |
| Providing historical context | **Supports** | Claude can provide background information (timeline of events, explanation of alliances) that students need before they can think critically. This is surface-level knowledge provision that frees class time for deeper thinking. |
| Evaluating the quality of historical reasoning | **Neutral to undermines** | Claude can assess whether an argument has a clear claim and evidence, but cannot reliably assess the QUALITY of historical reasoning — whether the student's use of evidence is genuinely historiographical or merely competent-sounding. The teacher must do this. |
| Developing intellectual humility about historical knowledge | **Undermines** | Claude presents information with confident fluency that may model certainty rather than the appropriate uncertainty that characterises genuine historical inquiry. If students see Claude as an authority rather than a tool to be evaluated, it undermines the disposition of intellectual humility the task is designed to cultivate. |

**Critical recommendation:**
Claude must be positioned as a tool students use AFTER they have done the thinking, not BEFORE or INSTEAD OF. The task sequence should be: (1) students engage with primary sources and construct their own argument without AI, (2) students then use Claude to challenge, extend, or refine their argument — asking it for counter-evidence, alternative interpretations, or structural feedback on their reasoning. This preserves the thinking demand while using AI productively. If the sequence is reversed — students ask Claude for an argument and then edit it — the historical thinking has not occurred.

### 4. AI-Specific Guidance

**Reliability for this domain:**
Claude performs well on synthesising established historiographical debates and providing factual overviews of WWI causes. It performs poorly on specific primary source quotations (may fabricate or misattribute), precise dates (may be slightly wrong), and nuanced historiographical distinctions (may flatten debates into balanced summaries). For this task, the teacher should treat Claude's factual claims about specific sources as unverified until checked, while treating its structural and analytical scaffolding as generally reliable.

**Verification requirements:**
Before the unit begins, the teacher should:
1. Ask Claude to summarise the Fischer thesis and the revisionist position. Check: does Claude represent Fischer's argument as genuinely provocative, or does it smooth it into a diplomatic summary? If the latter, the teacher needs to supplement with their own framing.
2. Ask Claude to quote from the key primary sources being used in the task (Kaiser's telegram, War Council minutes). Check: are the quotations accurate or fabricated? If fabricated, students must NOT be told to ask Claude for quotations — they must work from the sources provided by the teacher.
3. Ask Claude to evaluate a primary source's provenance. Check: does Claude assess the source historiographically, or does it produce a generic evaluation? If generic, this confirms that source evaluation must not be outsourced to Claude.

**Student critical evaluation — first lesson modelling move:**
In the first lesson using Claude, the teacher should model critical evaluation live in front of students. Procedure:
1. Ask Claude, in front of the class: "Was Germany primarily responsible for World War One?"
2. Read Claude's response aloud.
3. Ask students: "What does this response get right? What does it oversimplify? What evidence does it use, and is that evidence specific or vague?"
4. Point out: "Claude gives a balanced answer. But Fischer's argument was NOT balanced — it was a deliberate provocation. A balanced answer to this question is actually a distortion of the historiographical debate. What does that tell us about how to use Claude for history?"
This models the critical stance students need: AI produces fluent answers, but fluent does not mean historically accurate or historiographically faithful.

**Autonomy-dependency risk:**
This risk is HIGH for this task. The entire learning outcome — developing historical thinking through evidence-based argumentation — requires students to do the thinking themselves. If students use Claude to construct their Fischer-vs-revisionist argument, they are performing the task without doing the historical thinking the task was designed to develop. The product (a written argument) will exist, but the learning (the ability to evaluate evidence, weigh competing interpretations, and construct a defended position) will not have occurred.

Mitigation: Structure the task in two phases. Phase 1 (no AI): students read the sources, draft their argument by hand, and submit a first draft. Phase 2 (AI available): students use Claude to challenge their draft — asking it for counter-evidence, requesting feedback on their reasoning, or exploring an alternative interpretation. The teacher assesses the first draft (which shows the student's own thinking) and the revision (which shows their ability to engage with challenges and refine their reasoning).

### 5. Equity and Ethics

**Access equity:**
If Claude is used during class time on school devices, access is equitable during the lesson. However, students with AI access at home may refine their arguments further outside class, creating a differential advantage. Mitigation: assess the in-class first draft (pre-AI) as the primary evidence of historical thinking. The AI-assisted revision demonstrates a different skill (engaging with feedback) and should be assessed separately.

**Data privacy:**
Students aged 14-15 using Claude interact with a third-party AI platform. Requirements:
- School must have institutional terms of use that cover student data, or students must use the tool through a school-provisioned account that does not retain conversation data.
- Students should not be required to input personal information, emotional disclosures, or identifying details into the AI.
- The school's data protection officer (or equivalent) should be informed that student work is being processed by a third-party AI system.
- In jurisdictions covered by GDPR or similar frameworks, parental notification or consent may be required for this age group using third-party AI tools. The teacher should verify this with their school before proceeding.

**Content-specific ethics:**
The WWI topic does not involve personal disclosure, but the teacher should be aware that students may use AI to explore content related to war, violence, and mass death. Claude's responses to such queries are generally appropriate and contextualised, but the teacher should preview any AI-mediated engagement with sensitive historical content before assigning it.

### 6. Dispositional TPACK

The teacher needs to develop and maintain a continuous evaluative stance toward AI in their classroom — not a one-time setup but an ongoing practice:

- **In every lesson:** Ask "are my students using Claude to think more deeply, or to avoid thinking?" Look for the difference between a student who asks Claude "what evidence could someone use against my argument?" (productive use — extending thinking) and a student who asks Claude "write an argument about the Fischer thesis" (replacement use — avoiding thinking).
- **Between lessons:** Evaluate whether AI-generated student work is distinguishable from student-generated work. If it is not, the task design may need revision — the thinking demand may not be visible enough in the product.
- **Resist the sunk cost:** If the AI integration is not serving the learning outcome, be willing to remove it mid-unit. The technology should serve the history, not the other way around. An AI-free lesson in which students debate the Fischer thesis using only primary sources and their own reasoning may produce more historical thinking than an AI-mediated lesson that produces more polished but less genuinely argued responses.
- **Maintain Selwyn's critical stance:** When colleagues or administrators advocate for more AI use because it is "innovative," the teacher should ask: "Does it help my Year 9 students develop better historical thinking than they would without it?" If the answer is not clearly yes, the innovation argument is insufficient.

### 7. TPACK Development Plan

**Before using Claude with students:**
1. **Verify Claude's reliability on WWI content.** Run the three verification tests described in section 4. Know where Claude is accurate and where it is unreliable BEFORE students interact with it.
2. **Design the two-phase task structure.** Write the task so that Phase 1 (no AI) captures the student's own historical thinking, and Phase 2 (AI available) adds value without replacing the thinking. Prepare the Phase 2 prompt suggestions for students: "Ask Claude to challenge your argument," "Ask Claude what evidence a revisionist historian would use against your position."
3. **Prepare the first-lesson modelling move.** Script the live demonstration of critical evaluation described in section 4. Practise it once so it feels natural, not staged.

**During first use:**
- **Observe student queries.** Walk the room and look at what students are typing into Claude. Are they asking thinking-extending questions or task-completing questions? If most students are asking Claude to write their argument, pause the class and redirect: "I want you to ask Claude to CHALLENGE your argument, not to WRITE it."
- **Be ready to adjust access.** If Phase 1 drafts are significantly weaker than Phase 2 products, the AI may be doing the thinking. Consider restricting AI access to specific query types rather than open-ended use.
- **Note one student interaction that worked and one that did not.** After the lesson, record both. This builds the experiential TPACK that the skill cannot provide in advance.

**After first cycle:**
- **Compare Phase 1 and Phase 2 products.** If the quality improvement from Phase 1 to Phase 2 is in the reasoning (better engagement with counter-evidence, more nuanced qualification), the AI is serving the learning outcome. If the quality improvement is only in fluency and polish, the AI is serving the product but not the learning.
- **Ask students directly:** "Did using Claude help you think more carefully about the Fischer debate, or did it give you answers you could have reached yourself?" Student self-report is imperfect but informative.
- **Decide whether to continue.** If the AI integration produced better historical thinking (not just better-written arguments), refine the task design and continue. If it produced dependency or fluent but unthinking responses, consider restricting AI to specific uses (counter-argument generation, feedback on drafts) or removing it from this task entirely.

planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe.Stars: 118kUpdated Claude CodeCodexVerifiedInstall
contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies.Stars: 108kUpdated Claude CodeCodexVerifiedInstall
meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要".Stars: 108kUpdated Claude CodeCodexVerifiedInstall